Update Cargo.toml

DNS-Overrides + STUN fixes + Bind_addr prio + Fetch for unix-socket + ME/DC Method Detection + Metrics impovements

.gitignore

@@ -19,7 +19,5 @@ target
 #  and can be added to the global gitignore or merged into this file.  For a more nuclear
 #  option (not recommended) you can uncomment the following to ignore the entire idea folder.
 #.idea/
-*.rs
-target
-Cargo.lock
-src
+
+proxy-secret

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "telemt"
-version = "3.0.14"
+version = "3.1.3"
 edition = "2024"
 
 [dependencies]

README.md

@@ -2,6 +2,8 @@
 
 **Telemt** is a fast, secure, and feature-rich server written in Rust: it fully implements the official Telegram proxy algo and adds many production-ready improvements such as connection pooling, replay protection, detailed statistics, masking from "prying" eyes
 
+[**Telemt Chat in Telegram**](https://t.me/telemtrs)
+
 ## NEWS and EMERGENCY
 ### ✈️ Telemt 3 is released!
 <table>
@@ -10,28 +12,18 @@
 
 ### 🇷🇺 RU
 
-#### Драфтинг LTS и текущие улучшения
+#### Релиз 3.0.15 — 25 февраля
 
-С 21 февраля мы начали подготовку LTS-версии.
+25 февраля мы выпустили версию **3.0.15**
 
-Мы внимательно анализируем весь доступный фидбек.  
-Наша цель — сделать LTS-кандидаты максимально стабильными, тщательно отлаженными и готовыми к long-run и highload production-сценариям.
+Мы предполагаем, что она станет завершающей версией поколения 3.0 и уже сейчас мы рассматриваем её как **LTS-кандидата** для версии **3.1.0**!
 
----
+После нескольких дней детального анализа особенностей работы Middle-End мы спроектировали и реализовали продуманный режим **ротации ME Writer**. Данный режим позволяет поддерживать стабильно высокую производительность в long-run сценариях без возникновения ошибок, связанных с некорректной конфигурацией прокси
 
-#### Улучшения от 23 февраля
-
-23 февраля были внесены улучшения производительности в режимах **DC** и **Middle-End (ME)**, с акцентом на обратный канал (путь клиент → DC / ME).
-
-Дополнительно реализован ряд изменений, направленных на повышение устойчивости системы:
-
-- Смягчение сетевой нестабильности  
-- Повышение устойчивости к десинхронизации криптографии  
-- Снижение дрейфа сессий при неблагоприятных условиях  
-- Улучшение обработки ошибок в edge-case транспортных сценариях  
+Будем рады вашему фидбеку и предложениям по улучшению — особенно в части **статистики** и **UX**
 
 Релиз:  
-[3.0.12](https://github.com/telemt/telemt/releases/tag/3.0.12)
+[3.0.15](https://github.com/telemt/telemt/releases/tag/3.0.15)
 
 ---
 
@@ -48,28 +40,18 @@
 
 ### 🇬🇧 EN
 
-#### LTS Drafting and Ongoing Improvements
+#### Release 3.0.15 — February 25
 
-Starting February 21, we began drafting the upcoming LTS version.
+On February 25, we released version **3.0.15**
 
-We are carefully reviewing and analyzing all available feedback.  
-The goal is to ensure that LTS candidates are максимально stable, thoroughly debugged, and ready for long-run and high-load production scenarios.
+We expect this to become the final release of the 3.0 generation and at this point, we already see it as a strong **LTS candidate** for the upcoming **3.1.0** release!
 
----
+After several days of deep analysis of Middle-End behavior, we designed and implemented a well-engineered **ME Writer rotation mode**. This mode enables sustained high throughput in long-run scenarios while preventing proxy misconfiguration errors
 
-#### February 23 Improvements
-
-On February 23, we introduced performance improvements for both **DC** and **Middle-End (ME)** modes, specifically optimizing the reverse channel (client → DC / ME data path).
-
-Additionally, we implemented a set of robustness enhancements designed to:
-
-- Mitigate network-related instability
-- Improve resilience against cryptographic desynchronization
-- Reduce session drift under adverse conditions
-- Improve error handling in edge-case transport scenarios
+We are looking forward to your feedback and improvement proposals — especially regarding **statistics** and **UX**
 
 Release:  
-[3.0.12](https://github.com/telemt/telemt/releases/tag/3.0.12)
+[3.0.15](https://github.com/telemt/telemt/releases/tag/3.0.15)
 
 ---
 
@@ -209,6 +191,8 @@ then Ctrl+X -> Y -> Enter to save
 
 **5.** In Shell type `systemctl enable telemt` - then telemt will start with system startup, after the network is up
 
+**6.** In Shell type `journalctl -u telemt -n -g "links" --no-pager -o cat | tac` - get the connection links
+
 ## Configuration
 ### Minimal Configuration for First Start
 ```toml

config.full.toml

@@ -0,0 +1,205 @@
+# Telemt full config with default values.
+# Examples are kept in comments after '#'.
+
+# Top-level legacy field.
+show_link = [] # example: "*" or ["alice", "bob"]
+# default_dc = 2 # example: default DC for unmapped non-standard DCs
+
+[general]
+fast_mode = true
+use_middle_proxy = false
+# ad_tag = "00000000000000000000000000000000" # example
+# proxy_secret_path = "proxy-secret" # example custom path
+# middle_proxy_nat_ip = "203.0.113.10" # example public NAT IP override
+middle_proxy_nat_probe = true
+# middle_proxy_nat_stun = "stun.l.google.com:19302" # example
+# middle_proxy_nat_stun_servers = [] # example: ["stun1.l.google.com:19302", "stun2.l.google.com:19302"]
+middle_proxy_pool_size = 8
+middle_proxy_warm_standby = 16
+me_keepalive_enabled = true
+me_keepalive_interval_secs = 25
+me_keepalive_jitter_secs = 5
+me_keepalive_payload_random = true
+crypto_pending_buffer = 262144
+max_client_frame = 16777216
+desync_all_full = false
+beobachten = true
+beobachten_minutes = 10
+beobachten_flush_secs = 15
+beobachten_file = "cache/beobachten.txt"
+hardswap = true
+me_warmup_stagger_enabled = true
+me_warmup_step_delay_ms = 500
+me_warmup_step_jitter_ms = 300
+me_reconnect_max_concurrent_per_dc = 8
+me_reconnect_backoff_base_ms = 500
+me_reconnect_backoff_cap_ms = 30000
+me_reconnect_fast_retry_count = 12
+stun_iface_mismatch_ignore = false
+unknown_dc_log_path = "unknown-dc.txt" # to disable: set to null
+log_level = "normal" # debug | verbose | normal | silent
+disable_colors = false
+fast_mode_min_tls_record = 0
+update_every = 300
+me_reinit_every_secs = 900
+me_hardswap_warmup_delay_min_ms = 1000
+me_hardswap_warmup_delay_max_ms = 2000
+me_hardswap_warmup_extra_passes = 3
+me_hardswap_warmup_pass_backoff_base_ms = 500
+me_config_stable_snapshots = 2
+me_config_apply_cooldown_secs = 300
+proxy_secret_stable_snapshots = 2
+proxy_secret_rotate_runtime = true
+proxy_secret_len_max = 256
+me_pool_drain_ttl_secs = 90
+me_pool_min_fresh_ratio = 0.8
+me_reinit_drain_timeout_secs = 120
+# Legacy compatibility fields used when update_every is omitted.
+proxy_secret_auto_reload_secs = 3600
+proxy_config_auto_reload_secs = 3600
+ntp_check = true
+ntp_servers = ["pool.ntp.org"] # example: ["pool.ntp.org", "time.cloudflare.com"]
+auto_degradation_enabled = true
+degradation_min_unavailable_dc_groups = 2
+
+[general.modes]
+classic = false
+secure = false
+tls = true
+
+[general.links]
+show ="*" # example: "*" or ["alice", "bob"]
+# public_host = "proxy.example.com" # example explicit host/IP for tg:// links
+# public_port = 443 # example explicit port for tg:// links
+
+[network]
+ipv4 = true
+ipv6 = false # set true to enable IPv6
+prefer = 4 # 4 or 6
+multipath = false
+stun_servers = [
+  "stun.l.google.com:5349",
+  "stun1.l.google.com:3478",
+  "stun.gmx.net:3478",
+  "stun.l.google.com:19302",
+  "stun.1und1.de:3478",
+  "stun1.l.google.com:19302",
+  "stun2.l.google.com:19302",
+  "stun3.l.google.com:19302",
+  "stun4.l.google.com:19302",
+  "stun.services.mozilla.com:3478",
+  "stun.stunprotocol.org:3478",
+  "stun.nextcloud.com:3478",
+  "stun.voip.eutelia.it:3478",
+]
+stun_tcp_fallback = true
+http_ip_detect_urls = ["https://ifconfig.me/ip", "https://api.ipify.org"]
+cache_public_ip_path = "cache/public_ip.txt"
+
+[server]
+port = 443
+listen_addr_ipv4 = "0.0.0.0"
+listen_addr_ipv6 = "::"
+# listen_unix_sock = "/var/run/telemt.sock" # example
+# listen_unix_sock_perm = "0660" # example unix socket mode
+# listen_tcp = true # example explicit override (auto-detected when omitted)
+proxy_protocol = false
+# metrics_port = 9090 # example
+metrics_whitelist = ["127.0.0.1/32", "::1/128"]
+# Example explicit listeners (default: omitted, auto-generated from listen_addr_*):
+# [[server.listeners]]
+# ip = "0.0.0.0"
+# announce = "proxy-v4.example.com"
+# # announce_ip = "203.0.113.10" # deprecated alias
+# proxy_protocol = false
+# reuse_allow = false
+#
+# [[server.listeners]]
+# ip = "::"
+# announce = "proxy-v6.example.com"
+# proxy_protocol = false
+# reuse_allow = false
+
+[timeouts]
+client_handshake = 15
+tg_connect = 10
+client_keepalive = 60
+client_ack = 300
+me_one_retry = 3
+me_one_timeout_ms = 1500
+
+[censorship]
+tls_domain = "petrovich.ru"
+# tls_domains = ["example.com", "cdn.example.net"] # Additional domains for EE links
+mask = true
+# mask_host = "www.google.com" # example, defaults to tls_domain when both mask_host/mask_unix_sock are unset
+# mask_unix_sock = "/var/run/nginx.sock" # example, mutually exclusive with mask_host
+mask_port = 443
+# mask_proxy_protocol = 0 # Send PROXY protocol header to mask_host: 0 = off, 1 = v1 (text), 2 = v2 (binary)
+fake_cert_len = 2048 # if tls_emulation=false and default value is used, loader may randomize this value at runtime
+tls_emulation = true
+tls_front_dir = "tlsfront"
+server_hello_delay_min_ms = 0
+server_hello_delay_max_ms = 0
+tls_new_session_tickets = 0
+tls_full_cert_ttl_secs = 90
+alpn_enforce = true
+
+[access]
+replay_check_len = 65536
+replay_window_secs = 1800
+ignore_time_skew = false
+
+[access.users]
+# format: "username" = "32_hex_chars_secret"
+hello = "00000000000000000000000000000000"
+# alice = "11111111111111111111111111111111" # example
+
+[access.user_max_tcp_conns]
+# alice = 100 # example
+
+[access.user_expirations]
+# alice = "2078-01-01T00:00:00Z" # example
+
+[access.user_data_quota]
+# hello = 10737418240 # example bytes
+# alice = 10737418240 # example bytes
+
+[access.user_max_unique_ips]
+# hello = 10 # example
+# alice = 100 # example
+
+# Default behavior if [[upstreams]] is omitted: loader injects one direct upstream.
+# Example explicit upstreams:
+# [[upstreams]]
+# type = "direct"
+# interface = "eth0"
+# bind_addresses = ["192.0.2.10"]
+# weight = 1
+# enabled = true
+# scopes = "*"
+#
+# [[upstreams]]
+# type = "socks4"
+# address = "198.51.100.20:1080"
+# interface = "eth0"
+# user_id = "telemt"
+# weight = 1
+# enabled = true
+# scopes = "*"
+#
+# [[upstreams]]
+# type = "socks5"
+# address = "198.51.100.30:1080"
+# interface = "eth0"
+# username = "proxy-user"
+# password = "proxy-pass"
+# weight = 1
+# enabled = true
+# scopes = "*"
+
+# === DC Address Overrides ===
+# [dc_overrides]
+# "201" = "149.154.175.50:443" # example
+# "202" = ["149.154.167.51:443", "149.154.175.100:443"] # example
+# "203" = "91.105.192.100:443" # loader auto-adds this one when omitted

config.toml

@@ -1,11 +1,11 @@
+### Telemt Based Config.toml
+# We believe that these settings are sufficient for most scenarios 
+# where cutting-egde methods and parameters or special solutions are not needed
+
 # === General Settings ===
 [general]
-fast_mode = true
-use_middle_proxy = true
+use_middle_proxy = false
 # ad_tag = "00000000000000000000000000000000"
-# Path to proxy-secret binary (auto-downloaded if missing).
-proxy_secret_path = "proxy-secret"
-# disable_colors = false  # Disable colored output in logs (useful for files/systemd)
 
 # === Log Level ===
 # Log level: debug | verbose | normal | silent
@@ -13,46 +13,6 @@ proxy_secret_path = "proxy-secret"
 # RUST_LOG env var takes absolute priority over all of these
 log_level = "normal"
 
-# === Middle Proxy - ME ===
-# Public IP override for ME KDF when behind NAT; leave unset to auto-detect.
-# middle_proxy_nat_ip = "203.0.113.10"
-# Enable STUN probing to discover public IP:port for ME.
-middle_proxy_nat_probe = true
-# Primary STUN server (host:port); defaults to Telegram STUN when empty.
-middle_proxy_nat_stun = "stun.l.google.com:19302"
-# Optional fallback STUN servers list.
-middle_proxy_nat_stun_servers = ["stun1.l.google.com:19302", "stun2.l.google.com:19302"]
-# Desired number of concurrent ME writers in pool.
-middle_proxy_pool_size = 8
-# Pre-initialized warm-standby ME connections kept idle.
-middle_proxy_warm_standby = 8
-# Ignore STUN/interface mismatch and keep ME enabled even if IP differs.
-stun_iface_mismatch_ignore = false
-# Keepalive padding frames - fl==4
-me_keepalive_enabled = true
-me_keepalive_interval_secs = 25           # Period between keepalives
-me_keepalive_jitter_secs = 5              # Jitter added to interval
-me_keepalive_payload_random = true        # Randomize 4-byte payload (vs zeros)
-# Stagger extra ME connections on warmup to de-phase lifecycles.
-me_warmup_stagger_enabled = true
-me_warmup_step_delay_ms = 500             # Base delay between extra connects
-me_warmup_step_jitter_ms = 300            # Jitter for warmup delay
-# Reconnect policy knobs.
-me_reconnect_max_concurrent_per_dc = 4    # Parallel reconnects per DC - EXPERIMENTAL! UNSTABLE!
-me_reconnect_backoff_base_ms = 500        # Backoff start
-me_reconnect_backoff_cap_ms = 30000       # Backoff cap
-me_reconnect_fast_retry_count = 11        # Quick retries before backoff
-update_every = 7200                       # Resolve the active updater interval for ME infrastructure refresh tasks.
-crypto_pending_buffer = 262144            # Max pending ciphertext buffer per client writer (bytes). Controls FakeTLS backpressure vs throughput.
-max_client_frame = 16777216               # Maximum allowed client MTProto frame size (bytes).
-desync_all_full = false                   # Emit full crypto-desync forensic logs for every event. When false, full forensic details are emitted once per key window.
-auto_degradation_enabled = true           # Enable auto-degradation from ME to Direct-DC.
-degradation_min_unavailable_dc_groups = 2 # Minimum unavailable ME DC groups before degrading.
-hardswap = true                           # Enable C-like hard-swap for ME pool generations. When true, Telemt prewarms a new generation and switches once full coverage is reached.
-me_pool_drain_ttl_secs = 90               # Drain-TTL in seconds for stale ME writers after endpoint map changes. During TTL, stale writers may be used only as fallback for new bindings.
-me_pool_min_fresh_ratio = 0.8             # Minimum desired-DC coverage ratio required before draining stale writers. Range: 0.0..=1.0.
-me_reinit_drain_timeout_secs = 120        # Drain timeout in seconds for stale ME writers after endpoint map changes. Set to 0 to keep stale writers draining indefinitely (no force-close).
-
 [general.modes]
 classic = false
 secure = false
@@ -65,93 +25,24 @@ show = "*"
 # public_host = "proxy.example.com"  # Host (IP or domain) for tg:// links
 # public_port = 443                  # Port for tg:// links (default: server.port)
 
-# === Network Parameters ===
-[network]
-# Enable/disable families: true/false/auto(None)
-ipv4 = true
-ipv6 = false # UNSTABLE WITH ME
-# prefer = 4 or 6
-prefer = 4
-multipath = false # EXPERIMENTAL!
-
 # === Server Binding ===
 [server]
 port = 443
-listen_addr_ipv4 = "0.0.0.0"
-listen_addr_ipv6 = "::"
-# listen_unix_sock = "/var/run/telemt.sock" # Unix socket
-# listen_unix_sock_perm = "0666" # Socket file permissions
 # proxy_protocol = false           # Enable if behind HAProxy/nginx with PROXY protocol
 # metrics_port = 9090
-# metrics_whitelist = ["127.0.0.1", "::1"]
+# metrics_whitelist = ["127.0.0.1", "::1", "0.0.0.0/0"]
 
 # Listen on multiple interfaces/IPs - IPv4
 [[server.listeners]]
 ip = "0.0.0.0"
 
-# Listen on multiple interfaces/IPs - IPv6
-[[server.listeners]]
-ip = "::"
-
-# === Timeouts (in seconds) ===
-[timeouts]
-client_handshake = 30
-tg_connect = 10
-client_keepalive = 60
-client_ack = 300
-# Quick ME reconnects for single-address DCs (count and per-attempt timeout, ms).
-me_one_retry = 12
-me_one_timeout_ms = 1200
-
 # === Anti-Censorship & Masking ===
 [censorship]
 tls_domain = "petrovich.ru"
-# tls_domains = ["example.com", "cdn.example.net"] # Additional domains for EE links
 mask = true
-mask_port = 443
-# mask_host = "petrovich.ru" # Defaults to tls_domain if not set
-# mask_unix_sock = "/var/run/nginx.sock" # Unix socket (mutually exclusive with mask_host)
-fake_cert_len = 2048
-# tls_emulation = false        # Fetch real cert lengths and emulate TLS records
-# tls_front_dir = "tlsfront"   # Cache directory for TLS emulation
-
-# === Access Control & Users ===
-[access]
-replay_check_len = 65536
-replay_window_secs = 1800
-ignore_time_skew = false
+tls_emulation = true        # Fetch real cert lengths and emulate TLS records
+tls_front_dir = "tlsfront"   # Cache directory for TLS emulation
 
 [access.users]
 # format: "username" = "32_hex_chars_secret"
 hello = "00000000000000000000000000000000"
-
-# [access.user_max_tcp_conns]
-# hello = 50
-
-# [access.user_max_unique_ips]
-# hello = 5
-
-# [access.user_data_quota]
-# hello = 1073741824 # 1 GB
-
-# [access.user_expirations]
-# format: username = "[year]-[month]-[day]T[hour]:[minute]:[second]Z" UTC
-# hello = "2027-01-01T00:00:00Z"
-
-# === Upstreams & Routing ===
-[[upstreams]]
-type = "direct"
-enabled = true
-weight = 10
-# interface = "192.168.1.100"           # Bind outgoing to specific IP or iface name
-# bind_addresses = ["192.168.1.100"]    # List for round-robin binding (family must match target)
-
-# [[upstreams]]
-# type = "socks5"
-# address = "127.0.0.1:1080"
-# enabled = false
-# weight = 1
-
-# === DC Address Overrides ===
-# [dc_overrides]
-# "203" = "91.105.192.100:443"

docker-compose.yml

@@ -1,5 +1,6 @@
 services:
   telemt:
+    image: ghcr.io/telemt/telemt:latest
     build: .
     container_name: telemt
     restart: unless-stopped

docs/TUNING.de.md

@@ -0,0 +1,219 @@
+# Telemt Tuning-Leitfaden: Middle-End und Upstreams
+
+Dieses Dokument beschreibt das aktuelle Laufzeitverhalten für Middle-End (ME) und Upstream-Routing basierend auf:
+- `src/config/types.rs`
+- `src/config/defaults.rs`
+- `src/config/load.rs`
+- `src/transport/upstream.rs`
+
+Die unten angegebenen `Default`-Werte sind Code-Defaults (bei fehlendem Schlüssel), nicht zwingend die Werte aus `config.full.toml`.
+
+## Middle-End-Parameter
+
+### 1) ME-Grundmodus, NAT und STUN
+
+| Parameter | Typ | Default | Einschränkungen / Validierung | Laufzeiteffekt | Beispiel |
+|---|---|---:|---|---|---|
+| `general.use_middle_proxy` | `bool` | `true` | keine | Aktiviert den ME-Transportmodus. Bei `false` wird Direct-Modus verwendet. | `use_middle_proxy = true` |
+| `general.proxy_secret_path` | `Option<String>` | `"proxy-secret"` | Pfad kann `null` sein | Pfad zur Telegram-Infrastrukturdatei `proxy-secret`. | `proxy_secret_path = "proxy-secret"` |
+| `general.middle_proxy_nat_ip` | `Option<IpAddr>` | `null` | gültige IP bei gesetztem Wert | Manueller Override der öffentlichen NAT-IP für ME-Adressmaterial. | `middle_proxy_nat_ip = "203.0.113.10"` |
+| `general.middle_proxy_nat_probe` | `bool` | `true` | wird auf `true` erzwungen, wenn `use_middle_proxy=true` | Aktiviert NAT-Probing für ME. | `middle_proxy_nat_probe = true` |
+| `general.stun_nat_probe_concurrency` | `usize` | `8` | muss `> 0` sein | Maximale parallele STUN-Probes während NAT-Erkennung. | `stun_nat_probe_concurrency = 16` |
+| `network.stun_use` | `bool` | `true` | keine | Globaler STUN-Schalter. Bei `false` wird STUN deaktiviert. | `stun_use = true` |
+| `network.stun_servers` | `Vec<String>` | integrierter öffentlicher Pool | Duplikate/leer werden entfernt | Primäre STUN-Serverliste für NAT/Public-Endpoint-Erkennung. | `stun_servers = ["stun1.l.google.com:19302"]` |
+| `network.stun_tcp_fallback` | `bool` | `true` | keine | Aktiviert TCP-Fallback, wenn UDP-STUN blockiert ist. | `stun_tcp_fallback = true` |
+| `network.http_ip_detect_urls` | `Vec<String>` | `ifconfig.me` + `api.ipify.org` | keine | HTTP-Fallback zur öffentlichen IPv4-Erkennung, falls STUN ausfällt. | `http_ip_detect_urls = ["https://api.ipify.org"]` |
+| `general.stun_iface_mismatch_ignore` | `bool` | `false` | keine | Reserviertes Feld in der aktuellen Revision (derzeit kein aktiver Runtime-Verbrauch). | `stun_iface_mismatch_ignore = false` |
+| `timeouts.me_one_retry` | `u8` | `12` | keine | Anzahl schneller Reconnect-Versuche bei Single-Endpoint-DC-Fällen. | `me_one_retry = 6` |
+| `timeouts.me_one_timeout_ms` | `u64` | `1200` | keine | Timeout pro schnellem Einzelversuch (ms). | `me_one_timeout_ms = 1500` |
+
+### 2) Poolgröße, Keepalive und Reconnect-Policy
+
+| Parameter | Typ | Default | Einschränkungen / Validierung | Laufzeiteffekt | Beispiel |
+|---|---|---:|---|---|---|
+| `general.middle_proxy_pool_size` | `usize` | `8` | keine | Zielgröße des aktiven ME-Writer-Pools. | `middle_proxy_pool_size = 12` |
+| `general.middle_proxy_warm_standby` | `usize` | `16` | keine | Reserviertes Kompatibilitätsfeld in der aktuellen Revision (kein aktiver Runtime-Consumer). | `middle_proxy_warm_standby = 16` |
+| `general.me_keepalive_enabled` | `bool` | `true` | keine | Aktiviert periodischen ME-Keepalive/Ping-Traffic. | `me_keepalive_enabled = true` |
+| `general.me_keepalive_interval_secs` | `u64` | `25` | keine | Basisintervall für Keepalive (Sekunden). | `me_keepalive_interval_secs = 20` |
+| `general.me_keepalive_jitter_secs` | `u64` | `5` | keine | Keepalive-Jitter zur Vermeidung synchroner Peaks. | `me_keepalive_jitter_secs = 3` |
+| `general.me_keepalive_payload_random` | `bool` | `true` | keine | Randomisiert Keepalive-Payload-Bytes. | `me_keepalive_payload_random = true` |
+| `general.me_warmup_stagger_enabled` | `bool` | `true` | keine | Aktiviert gestaffeltes Warmup zusätzlicher ME-Verbindungen. | `me_warmup_stagger_enabled = true` |
+| `general.me_warmup_step_delay_ms` | `u64` | `500` | keine | Basisverzögerung zwischen Warmup-Schritten (ms). | `me_warmup_step_delay_ms = 300` |
+| `general.me_warmup_step_jitter_ms` | `u64` | `300` | keine | Zusätzlicher zufälliger Warmup-Jitter (ms). | `me_warmup_step_jitter_ms = 200` |
+| `general.me_reconnect_max_concurrent_per_dc` | `u32` | `8` | keine | Begrenzung paralleler Reconnect-Worker pro DC. | `me_reconnect_max_concurrent_per_dc = 12` |
+| `general.me_reconnect_backoff_base_ms` | `u64` | `500` | keine | Initiales Reconnect-Backoff (ms). | `me_reconnect_backoff_base_ms = 250` |
+| `general.me_reconnect_backoff_cap_ms` | `u64` | `30000` | keine | Maximales Reconnect-Backoff (ms). | `me_reconnect_backoff_cap_ms = 10000` |
+| `general.me_reconnect_fast_retry_count` | `u32` | `16` | keine | Budget für Sofort-Retries vor längerem Backoff. | `me_reconnect_fast_retry_count = 8` |
+
+### 3) Reinit/Hardswap, Secret-Rotation und Degradation
+
+| Parameter | Typ | Default | Einschränkungen / Validierung | Laufzeiteffekt | Beispiel |
+|---|---|---:|---|---|---|
+| `general.hardswap` | `bool` | `true` | keine | Aktiviert generation-basierte Hardswap-Strategie für den ME-Pool. | `hardswap = true` |
+| `general.me_reinit_every_secs` | `u64` | `900` | muss `> 0` sein | Intervall für periodische ME-Reinitialisierung. | `me_reinit_every_secs = 600` |
+| `general.me_hardswap_warmup_delay_min_ms` | `u64` | `1000` | muss `<= me_hardswap_warmup_delay_max_ms` sein | Untere Grenze für Warmup-Dial-Abstände. | `me_hardswap_warmup_delay_min_ms = 500` |
+| `general.me_hardswap_warmup_delay_max_ms` | `u64` | `2000` | muss `> 0` sein | Obere Grenze für Warmup-Dial-Abstände. | `me_hardswap_warmup_delay_max_ms = 1200` |
+| `general.me_hardswap_warmup_extra_passes` | `u8` | `3` | Bereich `[0,10]` | Zusätzliche Warmup-Pässe nach dem Basispass. | `me_hardswap_warmup_extra_passes = 2` |
+| `general.me_hardswap_warmup_pass_backoff_base_ms` | `u64` | `500` | muss `> 0` sein | Basis-Backoff zwischen zusätzlichen Warmup-Pässen. | `me_hardswap_warmup_pass_backoff_base_ms = 400` |
+| `general.me_config_stable_snapshots` | `u8` | `2` | muss `> 0` sein | Anzahl identischer ME-Config-Snapshots vor Apply. | `me_config_stable_snapshots = 3` |
+| `general.me_config_apply_cooldown_secs` | `u64` | `300` | keine | Cooldown zwischen angewendeten ME-Map-Updates. | `me_config_apply_cooldown_secs = 120` |
+| `general.proxy_secret_stable_snapshots` | `u8` | `2` | muss `> 0` sein | Anzahl identischer Secret-Snapshots vor Rotation. | `proxy_secret_stable_snapshots = 3` |
+| `general.proxy_secret_rotate_runtime` | `bool` | `true` | keine | Aktiviert Runtime-Rotation des Proxy-Secrets. | `proxy_secret_rotate_runtime = true` |
+| `general.proxy_secret_len_max` | `usize` | `256` | Bereich `[32,4096]` | Obergrenze für akzeptierte Secret-Länge. | `proxy_secret_len_max = 512` |
+| `general.update_every` | `Option<u64>` | `300` | wenn gesetzt: `> 0`; bei `null`: Legacy-Min-Fallback | Einheitliches Refresh-Intervall für ME-Config + Secret-Updater. | `update_every = 300` |
+| `general.me_pool_drain_ttl_secs` | `u64` | `90` | keine | Zeitraum, in dem stale Writer noch als Fallback zulässig sind. | `me_pool_drain_ttl_secs = 120` |
+| `general.me_pool_min_fresh_ratio` | `f32` | `0.8` | Bereich `[0.0,1.0]` | Coverage-Schwelle vor Drain der alten Generation. | `me_pool_min_fresh_ratio = 0.9` |
+| `general.me_reinit_drain_timeout_secs` | `u64` | `120` | `0` = kein Force-Close; wenn `>0 && < TTL`, dann auf TTL angehoben | Force-Close-Timeout für draining stale Writer. | `me_reinit_drain_timeout_secs = 0` |
+| `general.auto_degradation_enabled` | `bool` | `true` | keine | Reserviertes Kompatibilitätsfeld in aktueller Revision (kein aktiver Runtime-Consumer). | `auto_degradation_enabled = true` |
+| `general.degradation_min_unavailable_dc_groups` | `u8` | `2` | keine | Reservierter Kompatibilitäts-Schwellenwert in aktueller Revision (kein aktiver Runtime-Consumer). | `degradation_min_unavailable_dc_groups = 2` |
+
+## Deprecated / Legacy Parameter
+
+| Parameter | Status | Ersatz | Aktuelles Verhalten | Migrationshinweis |
+|---|---|---|---|---|
+| `general.middle_proxy_nat_stun` | Deprecated | `network.stun_servers` | Wird nur dann in `network.stun_servers` gemerged, wenn `network.stun_servers` nicht explizit gesetzt ist. | Wert nach `network.stun_servers` verschieben, Legacy-Key entfernen. |
+| `general.middle_proxy_nat_stun_servers` | Deprecated | `network.stun_servers` | Wird nur dann in `network.stun_servers` gemerged, wenn `network.stun_servers` nicht explizit gesetzt ist. | Werte nach `network.stun_servers` verschieben, Legacy-Key entfernen. |
+| `general.proxy_secret_auto_reload_secs` | Deprecated | `general.update_every` | Nur aktiv, wenn `update_every = null` (Legacy-Fallback). | `general.update_every` explizit setzen, Legacy-Key entfernen. |
+| `general.proxy_config_auto_reload_secs` | Deprecated | `general.update_every` | Nur aktiv, wenn `update_every = null` (Legacy-Fallback). | `general.update_every` explizit setzen, Legacy-Key entfernen. |
+
+## Wie Upstreams konfiguriert werden
+
+### Upstream-Schema
+
+| Feld | Gilt für | Typ | Pflicht | Default | Bedeutung |
+|---|---|---|---|---|---|
+| `[[upstreams]].type` | alle Upstreams | `"direct" \| "socks4" \| "socks5"` | ja | n/a | Upstream-Transporttyp. |
+| `[[upstreams]].weight` | alle Upstreams | `u16` | nein | `1` | Basisgewicht für weighted-random Auswahl. |
+| `[[upstreams]].enabled` | alle Upstreams | `bool` | nein | `true` | Deaktivierte Einträge werden beim Start ignoriert. |
+| `[[upstreams]].scopes` | alle Upstreams | `String` | nein | `""` | Komma-separierte Scope-Tags für Request-Routing. |
+| `interface` | `direct` | `Option<String>` | nein | `null` | Interface-Name (z. B. `eth0`) oder lokale Literal-IP. |
+| `bind_addresses` | `direct` | `Option<Vec<IpAddr>>` | nein | `null` | Explizite Source-IP-Kandidaten (strikter Vorrang vor `interface`). |
+| `address` | `socks4` | `String` | ja | n/a | SOCKS4-Server (`ip:port` oder `host:port`). |
+| `interface` | `socks4` | `Option<String>` | nein | `null` | Wird nur genutzt, wenn `address` als `ip:port` angegeben ist. |
+| `user_id` | `socks4` | `Option<String>` | nein | `null` | SOCKS4 User-ID für CONNECT. |
+| `address` | `socks5` | `String` | ja | n/a | SOCKS5-Server (`ip:port` oder `host:port`). |
+| `interface` | `socks5` | `Option<String>` | nein | `null` | Wird nur genutzt, wenn `address` als `ip:port` angegeben ist. |
+| `username` | `socks5` | `Option<String>` | nein | `null` | SOCKS5 Benutzername. |
+| `password` | `socks5` | `Option<String>` | nein | `null` | SOCKS5 Passwort. |
+
+### Runtime-Regeln (wichtig)
+
+1. Wenn `[[upstreams]]` fehlt, injiziert der Loader einen Default-`direct`-Upstream.
+2. Scope-Filterung basiert auf exaktem Token-Match:
+- mit Request-Scope -> nur Einträge, deren `scopes` genau dieses Token enthält;
+- ohne Request-Scope -> nur Einträge mit leerem `scopes`.
+3. Unter healthy Upstreams erfolgt die Auswahl per weighted random: `weight * latency_factor`.
+4. Gibt es im gefilterten Set keinen healthy Upstream, wird zufällig aus dem gefilterten Set gewählt.
+5. `direct`-Bind-Auflösung:
+- zuerst `bind_addresses` (nur gleiche IP-Familie wie Target);
+- bei `interface` (Name) + `bind_addresses` wird jede Candidate-IP gegen Interface-Adressen validiert;
+- ungültige Kandidaten werden mit `WARN` verworfen;
+- bleiben keine gültigen Kandidaten übrig, erfolgt unbound direct connect (`bind_ip=None`);
+- wenn `bind_addresses` nicht passt, wird `interface` verwendet (Literal-IP oder Interface-Primäradresse).
+6. Für `socks4/socks5` mit Hostname-`address` ist Interface-Binding nicht unterstützt und wird mit Warnung ignoriert.
+7. Runtime DNS Overrides werden für Hostname-Auflösung bei Upstream-Verbindungen genutzt.
+8. Im ME-Modus wird der gewählte Upstream auch für den ME-TCP-Dial-Pfad verwendet.
+9. Im ME-Modus ist bei `direct` mit bind/interface die STUN-Reflection bind-aware für KDF-Adressmaterial.
+10. Im ME-Modus werden bei SOCKS-Upstream `BND.ADDR/BND.PORT` für KDF verwendet, wenn gültig/öffentlich und gleiche IP-Familie.
+
+## Upstream-Konfigurationsbeispiele
+
+### Beispiel 1: Minimaler direct Upstream
+
+```toml
+[[upstreams]]
+type = "direct"
+weight = 1
+enabled = true
+```
+
+### Beispiel 2: direct mit Interface + expliziten bind IPs
+
+```toml
+[[upstreams]]
+type = "direct"
+interface = "eth0"
+bind_addresses = ["192.168.1.100", "192.168.1.101"]
+weight = 3
+enabled = true
+```
+
+### Beispiel 3: SOCKS5 Upstream mit Authentifizierung
+
+```toml
+[[upstreams]]
+type = "socks5"
+address = "198.51.100.30:1080"
+username = "proxy-user"
+password = "proxy-pass"
+weight = 2
+enabled = true
+```
+
+### Beispiel 4: Gemischte Upstreams mit Scopes
+
+```toml
+[[upstreams]]
+type = "direct"
+weight = 5
+enabled = true
+scopes = ""
+
+[[upstreams]]
+type = "socks5"
+address = "203.0.113.40:1080"
+username = "edge"
+password = "edgepass"
+weight = 3
+enabled = true
+scopes = "premium,me"
+```
+
+### Beispiel 5: ME-orientiertes Tuning-Profil
+
+```toml
+[general]
+use_middle_proxy = true
+proxy_secret_path = "proxy-secret"
+middle_proxy_nat_probe = true
+stun_nat_probe_concurrency = 16
+middle_proxy_pool_size = 12
+me_keepalive_enabled = true
+me_keepalive_interval_secs = 20
+me_keepalive_jitter_secs = 4
+me_reconnect_max_concurrent_per_dc = 12
+me_reconnect_backoff_base_ms = 300
+me_reconnect_backoff_cap_ms = 10000
+me_reconnect_fast_retry_count = 10
+hardswap = true
+me_reinit_every_secs = 600
+me_hardswap_warmup_delay_min_ms = 500
+me_hardswap_warmup_delay_max_ms = 1200
+me_hardswap_warmup_extra_passes = 2
+me_hardswap_warmup_pass_backoff_base_ms = 400
+me_config_stable_snapshots = 3
+me_config_apply_cooldown_secs = 120
+proxy_secret_stable_snapshots = 3
+proxy_secret_rotate_runtime = true
+proxy_secret_len_max = 512
+update_every = 300
+me_pool_drain_ttl_secs = 120
+me_pool_min_fresh_ratio = 0.9
+me_reinit_drain_timeout_secs = 180
+
+[timeouts]
+me_one_retry = 8
+me_one_timeout_ms = 1200
+
+[network]
+stun_use = true
+stun_tcp_fallback = true
+stun_servers = [
+  "stun1.l.google.com:19302",
+  "stun2.l.google.com:19302"
+]
+http_ip_detect_urls = [
+  "https://api.ipify.org",
+  "https://ifconfig.me/ip"
+]
+```

docs/TUNING.en.md

@@ -0,0 +1,219 @@
+# Telemt Tuning Guide: Middle-End and Upstreams
+
+This document describes the current runtime behavior for Middle-End (ME) and upstream routing based on:
+- `src/config/types.rs`
+- `src/config/defaults.rs`
+- `src/config/load.rs`
+- `src/transport/upstream.rs`
+
+Defaults below are code defaults (used when a key is omitted), not necessarily values from `config.full.toml` examples.
+
+## Middle-End Parameters
+
+### 1) Core ME mode, NAT, and STUN
+
+| Parameter | Type | Default | Constraints / validation | Runtime effect | Example |
+|---|---|---:|---|---|---|
+| `general.use_middle_proxy` | `bool` | `true` | none | Enables ME transport mode. If `false`, Direct mode is used. | `use_middle_proxy = true` |
+| `general.proxy_secret_path` | `Option<String>` | `"proxy-secret"` | path may be `null` | Path to Telegram infrastructure proxy-secret file. | `proxy_secret_path = "proxy-secret"` |
+| `general.middle_proxy_nat_ip` | `Option<IpAddr>` | `null` | valid IP when set | Manual public NAT IP override for ME address material. | `middle_proxy_nat_ip = "203.0.113.10"` |
+| `general.middle_proxy_nat_probe` | `bool` | `true` | auto-forced to `true` when `use_middle_proxy=true` | Enables ME NAT probing. | `middle_proxy_nat_probe = true` |
+| `general.stun_nat_probe_concurrency` | `usize` | `8` | must be `> 0` | Max parallel STUN probes during NAT discovery. | `stun_nat_probe_concurrency = 16` |
+| `network.stun_use` | `bool` | `true` | none | Global STUN switch. If `false`, STUN probing is disabled. | `stun_use = true` |
+| `network.stun_servers` | `Vec<String>` | built-in public pool | deduplicated + empty values removed | Primary STUN server list for NAT/public endpoint discovery. | `stun_servers = ["stun1.l.google.com:19302"]` |
+| `network.stun_tcp_fallback` | `bool` | `true` | none | Enables TCP fallback path when UDP STUN is blocked. | `stun_tcp_fallback = true` |
+| `network.http_ip_detect_urls` | `Vec<String>` | `ifconfig.me` + `api.ipify.org` | none | HTTP fallback for public IPv4 detection if STUN is unavailable. | `http_ip_detect_urls = ["https://api.ipify.org"]` |
+| `general.stun_iface_mismatch_ignore` | `bool` | `false` | none | Reserved flag in current revision (not consumed by runtime path). | `stun_iface_mismatch_ignore = false` |
+| `timeouts.me_one_retry` | `u8` | `12` | none | Fast reconnect attempts for single-endpoint DC cases. | `me_one_retry = 6` |
+| `timeouts.me_one_timeout_ms` | `u64` | `1200` | none | Timeout per quick single-endpoint attempt (ms). | `me_one_timeout_ms = 1500` |
+
+### 2) Pool size, keepalive, and reconnect policy
+
+| Parameter | Type | Default | Constraints / validation | Runtime effect | Example |
+|---|---|---:|---|---|---|
+| `general.middle_proxy_pool_size` | `usize` | `8` | none | Target active ME writer pool size. | `middle_proxy_pool_size = 12` |
+| `general.middle_proxy_warm_standby` | `usize` | `16` | none | Reserved compatibility field in current revision (no active runtime consumer). | `middle_proxy_warm_standby = 16` |
+| `general.me_keepalive_enabled` | `bool` | `true` | none | Enables periodic ME keepalive/ping traffic. | `me_keepalive_enabled = true` |
+| `general.me_keepalive_interval_secs` | `u64` | `25` | none | Base keepalive interval (seconds). | `me_keepalive_interval_secs = 20` |
+| `general.me_keepalive_jitter_secs` | `u64` | `5` | none | Keepalive jitter to avoid synchronization bursts. | `me_keepalive_jitter_secs = 3` |
+| `general.me_keepalive_payload_random` | `bool` | `true` | none | Randomizes keepalive payload bytes. | `me_keepalive_payload_random = true` |
+| `general.me_warmup_stagger_enabled` | `bool` | `true` | none | Staggers extra ME warmup dials to avoid spikes. | `me_warmup_stagger_enabled = true` |
+| `general.me_warmup_step_delay_ms` | `u64` | `500` | none | Base delay between warmup dial steps (ms). | `me_warmup_step_delay_ms = 300` |
+| `general.me_warmup_step_jitter_ms` | `u64` | `300` | none | Additional random delay for warmup steps (ms). | `me_warmup_step_jitter_ms = 200` |
+| `general.me_reconnect_max_concurrent_per_dc` | `u32` | `8` | none | Limits concurrent reconnect workers per DC in health recovery. | `me_reconnect_max_concurrent_per_dc = 12` |
+| `general.me_reconnect_backoff_base_ms` | `u64` | `500` | none | Initial reconnect backoff (ms). | `me_reconnect_backoff_base_ms = 250` |
+| `general.me_reconnect_backoff_cap_ms` | `u64` | `30000` | none | Maximum reconnect backoff (ms). | `me_reconnect_backoff_cap_ms = 10000` |
+| `general.me_reconnect_fast_retry_count` | `u32` | `16` | none | Immediate retry budget before long backoff behavior. | `me_reconnect_fast_retry_count = 8` |
+
+### 3) Reinit/hardswap, secret rotation, and degradation
+
+| Parameter | Type | Default | Constraints / validation | Runtime effect | Example |
+|---|---|---:|---|---|---|
+| `general.hardswap` | `bool` | `true` | none | Enables generation-based ME hardswap strategy. | `hardswap = true` |
+| `general.me_reinit_every_secs` | `u64` | `900` | must be `> 0` | Periodic ME reinit interval. | `me_reinit_every_secs = 600` |
+| `general.me_hardswap_warmup_delay_min_ms` | `u64` | `1000` | must be `<= me_hardswap_warmup_delay_max_ms` | Lower bound for hardswap warmup dial spacing. | `me_hardswap_warmup_delay_min_ms = 500` |
+| `general.me_hardswap_warmup_delay_max_ms` | `u64` | `2000` | must be `> 0` | Upper bound for hardswap warmup dial spacing. | `me_hardswap_warmup_delay_max_ms = 1200` |
+| `general.me_hardswap_warmup_extra_passes` | `u8` | `3` | must be within `[0,10]` | Additional warmup passes after base pass. | `me_hardswap_warmup_extra_passes = 2` |
+| `general.me_hardswap_warmup_pass_backoff_base_ms` | `u64` | `500` | must be `> 0` | Base backoff between extra warmup passes. | `me_hardswap_warmup_pass_backoff_base_ms = 400` |
+| `general.me_config_stable_snapshots` | `u8` | `2` | must be `> 0` | Number of identical ME config snapshots required before apply. | `me_config_stable_snapshots = 3` |
+| `general.me_config_apply_cooldown_secs` | `u64` | `300` | none | Cooldown between applied ME map updates. | `me_config_apply_cooldown_secs = 120` |
+| `general.proxy_secret_stable_snapshots` | `u8` | `2` | must be `> 0` | Number of identical proxy-secret snapshots required before rotation. | `proxy_secret_stable_snapshots = 3` |
+| `general.proxy_secret_rotate_runtime` | `bool` | `true` | none | Enables runtime proxy-secret rotation. | `proxy_secret_rotate_runtime = true` |
+| `general.proxy_secret_len_max` | `usize` | `256` | must be within `[32,4096]` | Upper limit for accepted proxy-secret length. | `proxy_secret_len_max = 512` |
+| `general.update_every` | `Option<u64>` | `300` | if set: must be `> 0`; if `null`: legacy min fallback | Unified refresh interval for ME config + secret updater. | `update_every = 300` |
+| `general.me_pool_drain_ttl_secs` | `u64` | `90` | none | Time window where stale writers remain fallback-eligible. | `me_pool_drain_ttl_secs = 120` |
+| `general.me_pool_min_fresh_ratio` | `f32` | `0.8` | must be within `[0.0,1.0]` | Coverage threshold before stale generation can be drained. | `me_pool_min_fresh_ratio = 0.9` |
+| `general.me_reinit_drain_timeout_secs` | `u64` | `120` | `0` means no force-close; if `>0 && < TTL` it is bumped to TTL | Force-close timeout for draining stale writers. | `me_reinit_drain_timeout_secs = 0` |
+| `general.auto_degradation_enabled` | `bool` | `true` | none | Reserved compatibility flag in current revision (no active runtime consumer). | `auto_degradation_enabled = true` |
+| `general.degradation_min_unavailable_dc_groups` | `u8` | `2` | none | Reserved compatibility threshold in current revision (no active runtime consumer). | `degradation_min_unavailable_dc_groups = 2` |
+
+## Deprecated / Legacy Parameters
+
+| Parameter | Status | Replacement | Current behavior | Migration recommendation |
+|---|---|---|---|---|
+| `general.middle_proxy_nat_stun` | Deprecated | `network.stun_servers` | Merged into `network.stun_servers` only when `network.stun_servers` is not explicitly set. | Move value into `network.stun_servers` and remove legacy key. |
+| `general.middle_proxy_nat_stun_servers` | Deprecated | `network.stun_servers` | Merged into `network.stun_servers` only when `network.stun_servers` is not explicitly set. | Move values into `network.stun_servers` and remove legacy key. |
+| `general.proxy_secret_auto_reload_secs` | Deprecated | `general.update_every` | Used only when `update_every = null` (legacy fallback path). | Set `general.update_every` explicitly and remove legacy key. |
+| `general.proxy_config_auto_reload_secs` | Deprecated | `general.update_every` | Used only when `update_every = null` (legacy fallback path). | Set `general.update_every` explicitly and remove legacy key. |
+
+## How Upstreams Are Configured
+
+### Upstream schema
+
+| Field | Applies to | Type | Required | Default | Meaning |
+|---|---|---|---|---|---|
+| `[[upstreams]].type` | all upstreams | `"direct" \| "socks4" \| "socks5"` | yes | n/a | Upstream transport type. |
+| `[[upstreams]].weight` | all upstreams | `u16` | no | `1` | Base weight for weighted-random selection. |
+| `[[upstreams]].enabled` | all upstreams | `bool` | no | `true` | Disabled entries are ignored at startup. |
+| `[[upstreams]].scopes` | all upstreams | `String` | no | `""` | Comma-separated scope tags for request-level routing. |
+| `interface` | `direct` | `Option<String>` | no | `null` | Interface name (e.g. `eth0`) or literal local IP for bind selection. |
+| `bind_addresses` | `direct` | `Option<Vec<IpAddr>>` | no | `null` | Explicit local source IP candidates (strict priority over `interface`). |
+| `address` | `socks4` | `String` | yes | n/a | SOCKS4 server endpoint (`ip:port` or `host:port`). |
+| `interface` | `socks4` | `Option<String>` | no | `null` | Used only for SOCKS server `ip:port` dial path. |
+| `user_id` | `socks4` | `Option<String>` | no | `null` | SOCKS4 user ID for CONNECT request. |
+| `address` | `socks5` | `String` | yes | n/a | SOCKS5 server endpoint (`ip:port` or `host:port`). |
+| `interface` | `socks5` | `Option<String>` | no | `null` | Used only for SOCKS server `ip:port` dial path. |
+| `username` | `socks5` | `Option<String>` | no | `null` | SOCKS5 username auth. |
+| `password` | `socks5` | `Option<String>` | no | `null` | SOCKS5 password auth. |
+
+### Runtime rules (important)
+
+1. If `[[upstreams]]` is omitted, loader injects one default `direct` upstream.
+2. Scope filtering is exact-token based:
+- when request scope is set -> only entries whose `scopes` contains that exact token;
+- when request scope is not set -> only entries with empty `scopes`.
+3. Healthy upstreams are selected by weighted random using: `weight * latency_factor`.
+4. If no healthy upstream exists in filtered set, random selection is used among filtered entries.
+5. `direct` bind resolution order:
+- `bind_addresses` candidates (same IP family as target) first;
+- if `interface` is an interface name and `bind_addresses` is set, each candidate IP is validated against addresses currently assigned to that interface;
+- invalid candidates are dropped with `WARN`;
+- if no valid candidate remains, connection falls back to unbound direct connect (`bind_ip=None`);
+- if no `bind_addresses` candidate, `interface` is used (literal IP or resolved interface primary IP).
+6. For `socks4/socks5` with `address` as hostname, interface binding is not supported and is ignored with warning.
+7. Runtime DNS overrides are used for upstream hostname resolution.
+8. In ME mode, the selected upstream is also used for ME TCP dial path.
+9. In ME mode for `direct` upstream with bind/interface, STUN reflection logic is bind-aware for KDF source material.
+10. In ME mode for SOCKS upstream, SOCKS `BND.ADDR/BND.PORT` is used for KDF when it is valid/public for the same family.
+
+## Upstream Configuration Examples
+
+### Example 1: Minimal direct upstream
+
+```toml
+[[upstreams]]
+type = "direct"
+weight = 1
+enabled = true
+```
+
+### Example 2: Direct with interface + explicit bind addresses
+
+```toml
+[[upstreams]]
+type = "direct"
+interface = "eth0"
+bind_addresses = ["192.168.1.100", "192.168.1.101"]
+weight = 3
+enabled = true
+```
+
+### Example 3: SOCKS5 upstream with authentication
+
+```toml
+[[upstreams]]
+type = "socks5"
+address = "198.51.100.30:1080"
+username = "proxy-user"
+password = "proxy-pass"
+weight = 2
+enabled = true
+```
+
+### Example 4: Mixed upstreams with scopes
+
+```toml
+[[upstreams]]
+type = "direct"
+weight = 5
+enabled = true
+scopes = ""
+
+[[upstreams]]
+type = "socks5"
+address = "203.0.113.40:1080"
+username = "edge"
+password = "edgepass"
+weight = 3
+enabled = true
+scopes = "premium,me"
+```
+
+### Example 5: ME-focused tuning profile
+
+```toml
+[general]
+use_middle_proxy = true
+proxy_secret_path = "proxy-secret"
+middle_proxy_nat_probe = true
+stun_nat_probe_concurrency = 16
+middle_proxy_pool_size = 12
+me_keepalive_enabled = true
+me_keepalive_interval_secs = 20
+me_keepalive_jitter_secs = 4
+me_reconnect_max_concurrent_per_dc = 12
+me_reconnect_backoff_base_ms = 300
+me_reconnect_backoff_cap_ms = 10000
+me_reconnect_fast_retry_count = 10
+hardswap = true
+me_reinit_every_secs = 600
+me_hardswap_warmup_delay_min_ms = 500
+me_hardswap_warmup_delay_max_ms = 1200
+me_hardswap_warmup_extra_passes = 2
+me_hardswap_warmup_pass_backoff_base_ms = 400
+me_config_stable_snapshots = 3
+me_config_apply_cooldown_secs = 120
+proxy_secret_stable_snapshots = 3
+proxy_secret_rotate_runtime = true
+proxy_secret_len_max = 512
+update_every = 300
+me_pool_drain_ttl_secs = 120
+me_pool_min_fresh_ratio = 0.9
+me_reinit_drain_timeout_secs = 180
+
+[timeouts]
+me_one_retry = 8
+me_one_timeout_ms = 1200
+
+[network]
+stun_use = true
+stun_tcp_fallback = true
+stun_servers = [
+  "stun1.l.google.com:19302",
+  "stun2.l.google.com:19302"
+]
+http_ip_detect_urls = [
+  "https://api.ipify.org",
+  "https://ifconfig.me/ip"
+]
+```

docs/TUNING.ru.md

@@ -0,0 +1,219 @@
+# Руководство по тюнингу Telemt: Middle-End и Upstreams
+
+Документ описывает актуальное поведение Middle-End (ME) и маршрутизации через upstream на основе:
+- `src/config/types.rs`
+- `src/config/defaults.rs`
+- `src/config/load.rs`
+- `src/transport/upstream.rs`
+
+Значения `Default` ниже — это значения из кода при отсутствии ключа в конфиге, а не обязательно значения из примеров `config.full.toml`.
+
+## Параметры Middle-End
+
+### 1) Базовый режим ME, NAT и STUN
+
+| Параметр | Тип | Default | Ограничения / валидация | Влияние на runtime | Пример |
+|---|---|---:|---|---|---|
+| `general.use_middle_proxy` | `bool` | `true` | нет | Включает транспорт ME. При `false` используется Direct-режим. | `use_middle_proxy = true` |
+| `general.proxy_secret_path` | `Option<String>` | `"proxy-secret"` | путь может быть `null` | Путь к инфраструктурному proxy-secret Telegram. | `proxy_secret_path = "proxy-secret"` |
+| `general.middle_proxy_nat_ip` | `Option<IpAddr>` | `null` | валидный IP при задании | Ручной override публичного NAT IP для адресного материала ME. | `middle_proxy_nat_ip = "203.0.113.10"` |
+| `general.middle_proxy_nat_probe` | `bool` | `true` | авто-принудительно `true`, если `use_middle_proxy=true` | Включает NAT probing для ME. | `middle_proxy_nat_probe = true` |
+| `general.stun_nat_probe_concurrency` | `usize` | `8` | должно быть `> 0` | Максимум параллельных STUN-проб при NAT-детекте. | `stun_nat_probe_concurrency = 16` |
+| `network.stun_use` | `bool` | `true` | нет | Глобальный переключатель STUN. При `false` STUN отключен. | `stun_use = true` |
+| `network.stun_servers` | `Vec<String>` | встроенный публичный пул | удаляются дубликаты и пустые значения | Основной список STUN-серверов для NAT/public endpoint discovery. | `stun_servers = ["stun1.l.google.com:19302"]` |
+| `network.stun_tcp_fallback` | `bool` | `true` | нет | Включает TCP fallback, если UDP STUN недоступен. | `stun_tcp_fallback = true` |
+| `network.http_ip_detect_urls` | `Vec<String>` | `ifconfig.me` + `api.ipify.org` | нет | HTTP fallback для определения публичного IPv4 при недоступности STUN. | `http_ip_detect_urls = ["https://api.ipify.org"]` |
+| `general.stun_iface_mismatch_ignore` | `bool` | `false` | нет | Зарезервированный флаг в текущей ревизии (runtime его не использует). | `stun_iface_mismatch_ignore = false` |
+| `timeouts.me_one_retry` | `u8` | `12` | нет | Количество быстрых reconnect-попыток для DC с одним endpoint. | `me_one_retry = 6` |
+| `timeouts.me_one_timeout_ms` | `u64` | `1200` | нет | Таймаут одной быстрой попытки (мс). | `me_one_timeout_ms = 1500` |
+
+### 2) Размер пула, keepalive и reconnect-политика
+
+| Параметр | Тип | Default | Ограничения / валидация | Влияние на runtime | Пример |
+|---|---|---:|---|---|---|
+| `general.middle_proxy_pool_size` | `usize` | `8` | нет | Целевой размер активного пула ME-writer соединений. | `middle_proxy_pool_size = 12` |
+| `general.middle_proxy_warm_standby` | `usize` | `16` | нет | Зарезервированное поле совместимости в текущей ревизии (активного runtime-consumer нет). | `middle_proxy_warm_standby = 16` |
+| `general.me_keepalive_enabled` | `bool` | `true` | нет | Включает периодические keepalive/ping кадры ME. | `me_keepalive_enabled = true` |
+| `general.me_keepalive_interval_secs` | `u64` | `25` | нет | Базовый интервал keepalive (сек). | `me_keepalive_interval_secs = 20` |
+| `general.me_keepalive_jitter_secs` | `u64` | `5` | нет | Джиттер keepalive для предотвращения синхронных всплесков. | `me_keepalive_jitter_secs = 3` |
+| `general.me_keepalive_payload_random` | `bool` | `true` | нет | Рандомизирует payload keepalive-кадров. | `me_keepalive_payload_random = true` |
+| `general.me_warmup_stagger_enabled` | `bool` | `true` | нет | Включает staggered warmup дополнительных ME-коннектов. | `me_warmup_stagger_enabled = true` |
+| `general.me_warmup_step_delay_ms` | `u64` | `500` | нет | Базовая задержка между шагами warmup (мс). | `me_warmup_step_delay_ms = 300` |
+| `general.me_warmup_step_jitter_ms` | `u64` | `300` | нет | Дополнительный случайный warmup-джиттер (мс). | `me_warmup_step_jitter_ms = 200` |
+| `general.me_reconnect_max_concurrent_per_dc` | `u32` | `8` | нет | Ограничивает параллельные reconnect worker'ы на один DC. | `me_reconnect_max_concurrent_per_dc = 12` |
+| `general.me_reconnect_backoff_base_ms` | `u64` | `500` | нет | Начальный backoff reconnect (мс). | `me_reconnect_backoff_base_ms = 250` |
+| `general.me_reconnect_backoff_cap_ms` | `u64` | `30000` | нет | Верхняя граница backoff reconnect (мс). | `me_reconnect_backoff_cap_ms = 10000` |
+| `general.me_reconnect_fast_retry_count` | `u32` | `16` | нет | Бюджет быстрых retry до длинного backoff. | `me_reconnect_fast_retry_count = 8` |
+
+### 3) Reinit/hardswap, ротация секрета и деградация
+
+| Параметр | Тип | Default | Ограничения / валидация | Влияние на runtime | Пример |
+|---|---|---:|---|---|---|
+| `general.hardswap` | `bool` | `true` | нет | Включает generation-based стратегию hardswap для ME-пула. | `hardswap = true` |
+| `general.me_reinit_every_secs` | `u64` | `900` | должно быть `> 0` | Интервал периодического reinit ME-пула. | `me_reinit_every_secs = 600` |
+| `general.me_hardswap_warmup_delay_min_ms` | `u64` | `1000` | должно быть `<= me_hardswap_warmup_delay_max_ms` | Нижняя граница пауз между warmup dial попытками. | `me_hardswap_warmup_delay_min_ms = 500` |
+| `general.me_hardswap_warmup_delay_max_ms` | `u64` | `2000` | должно быть `> 0` | Верхняя граница пауз между warmup dial попытками. | `me_hardswap_warmup_delay_max_ms = 1200` |
+| `general.me_hardswap_warmup_extra_passes` | `u8` | `3` | диапазон `[0,10]` | Дополнительные warmup-проходы после базового. | `me_hardswap_warmup_extra_passes = 2` |
+| `general.me_hardswap_warmup_pass_backoff_base_ms` | `u64` | `500` | должно быть `> 0` | Базовый backoff между extra-pass в warmup. | `me_hardswap_warmup_pass_backoff_base_ms = 400` |
+| `general.me_config_stable_snapshots` | `u8` | `2` | должно быть `> 0` | Количество одинаковых snapshot перед применением ME map update. | `me_config_stable_snapshots = 3` |
+| `general.me_config_apply_cooldown_secs` | `u64` | `300` | нет | Cooldown между применёнными обновлениями ME map. | `me_config_apply_cooldown_secs = 120` |
+| `general.proxy_secret_stable_snapshots` | `u8` | `2` | должно быть `> 0` | Количество одинаковых snapshot перед runtime-rotation proxy-secret. | `proxy_secret_stable_snapshots = 3` |
+| `general.proxy_secret_rotate_runtime` | `bool` | `true` | нет | Включает runtime-ротацию proxy-secret. | `proxy_secret_rotate_runtime = true` |
+| `general.proxy_secret_len_max` | `usize` | `256` | диапазон `[32,4096]` | Верхний лимит длины принимаемого proxy-secret. | `proxy_secret_len_max = 512` |
+| `general.update_every` | `Option<u64>` | `300` | если задано: `> 0`; если `null`: fallback на legacy минимум | Единый интервал refresh для ME config + secret updater. | `update_every = 300` |
+| `general.me_pool_drain_ttl_secs` | `u64` | `90` | нет | Время, когда stale writer ещё может использоваться как fallback. | `me_pool_drain_ttl_secs = 120` |
+| `general.me_pool_min_fresh_ratio` | `f32` | `0.8` | диапазон `[0.0,1.0]` | Порог покрытия fresh-поколения перед drain старого поколения. | `me_pool_min_fresh_ratio = 0.9` |
+| `general.me_reinit_drain_timeout_secs` | `u64` | `120` | `0` = без force-close; если `>0 && < TTL`, поднимается до TTL | Таймаут force-close для draining stale writer. | `me_reinit_drain_timeout_secs = 0` |
+| `general.auto_degradation_enabled` | `bool` | `true` | нет | Зарезервированный флаг совместимости в текущей ревизии (активного runtime-consumer нет). | `auto_degradation_enabled = true` |
+| `general.degradation_min_unavailable_dc_groups` | `u8` | `2` | нет | Зарезервированный порог совместимости в текущей ревизии (активного runtime-consumer нет). | `degradation_min_unavailable_dc_groups = 2` |
+
+## Устаревшие / legacy параметры
+
+| Параметр | Статус | Замена | Текущее поведение | Рекомендация миграции |
+|---|---|---|---|---|
+| `general.middle_proxy_nat_stun` | Deprecated | `network.stun_servers` | Добавляется в `network.stun_servers`, только если `network.stun_servers` не задан явно. | Перенести значение в `network.stun_servers`, legacy-ключ удалить. |
+| `general.middle_proxy_nat_stun_servers` | Deprecated | `network.stun_servers` | Добавляется в `network.stun_servers`, только если `network.stun_servers` не задан явно. | Перенести значения в `network.stun_servers`, legacy-ключ удалить. |
+| `general.proxy_secret_auto_reload_secs` | Deprecated | `general.update_every` | Используется только если `update_every = null` (legacy fallback). | Явно задать `general.update_every`, legacy-ключ удалить. |
+| `general.proxy_config_auto_reload_secs` | Deprecated | `general.update_every` | Используется только если `update_every = null` (legacy fallback). | Явно задать `general.update_every`, legacy-ключ удалить. |
+
+## Как конфигурируются Upstreams
+
+### Схема upstream
+
+| Поле | Применимость | Тип | Обязательно | Default | Назначение |
+|---|---|---|---|---|---|
+| `[[upstreams]].type` | все upstream | `"direct" \| "socks4" \| "socks5"` | да | n/a | Тип upstream транспорта. |
+| `[[upstreams]].weight` | все upstream | `u16` | нет | `1` | Базовый вес в weighted-random выборе. |
+| `[[upstreams]].enabled` | все upstream | `bool` | нет | `true` | Выключенные записи игнорируются на старте. |
+| `[[upstreams]].scopes` | все upstream | `String` | нет | `""` | Список scope-токенов через запятую для маршрутизации. |
+| `interface` | `direct` | `Option<String>` | нет | `null` | Имя интерфейса (например `eth0`) или literal локальный IP. |
+| `bind_addresses` | `direct` | `Option<Vec<IpAddr>>` | нет | `null` | Явные кандидаты source IP (имеют приоритет над `interface`). |
+| `address` | `socks4` | `String` | да | n/a | Адрес SOCKS4 сервера (`ip:port` или `host:port`). |
+| `interface` | `socks4` | `Option<String>` | нет | `null` | Используется только если `address` задан как `ip:port`. |
+| `user_id` | `socks4` | `Option<String>` | нет | `null` | SOCKS4 user ID в CONNECT-запросе. |
+| `address` | `socks5` | `String` | да | n/a | Адрес SOCKS5 сервера (`ip:port` или `host:port`). |
+| `interface` | `socks5` | `Option<String>` | нет | `null` | Используется только если `address` задан как `ip:port`. |
+| `username` | `socks5` | `Option<String>` | нет | `null` | Логин SOCKS5 auth. |
+| `password` | `socks5` | `Option<String>` | нет | `null` | Пароль SOCKS5 auth. |
+
+### Runtime-правила
+
+1. Если `[[upstreams]]` отсутствует, loader добавляет один upstream `direct` по умолчанию.
+2. Scope-фильтрация — по точному совпадению токена:
+- если scope запроса задан -> используются только записи, где `scopes` содержит такой же токен;
+- если scope запроса не задан -> используются только записи с пустым `scopes`.
+3. Среди healthy upstream используется weighted-random выбор: `weight * latency_factor`.
+4. Если в отфильтрованном наборе нет healthy upstream, выбирается случайный из отфильтрованных.
+5. Порядок выбора bind для `direct`:
+- сначала `bind_addresses` (только IP нужного семейства);
+- если одновременно заданы `interface` (имя) и `bind_addresses`, каждый IP проверяется на принадлежность интерфейсу;
+- несовпадающие IP отбрасываются с `WARN`;
+- если валидных IP не осталось, используется unbound direct connect (`bind_ip=None`);
+- если `bind_addresses` не подходит, применяется `interface` (literal IP или адрес интерфейса).
+6. Для `socks4/socks5` с `address` в виде hostname интерфейсный bind не поддерживается и игнорируется с предупреждением.
+7. Runtime DNS overrides применяются к резолвингу hostname в upstream-подключениях.
+8. В ME-режиме выбранный upstream также используется для ME TCP dial path.
+9. В ME-режиме для `direct` upstream с bind/interface STUN-рефлексия выполняется bind-aware для KDF материала.
+10. В ME-режиме для SOCKS upstream используются `BND.ADDR/BND.PORT` для KDF, если адрес валиден/публичен и соответствует IP family.
+
+## Примеры конфигурации Upstreams
+
+### Пример 1: минимальный direct upstream
+
+```toml
+[[upstreams]]
+type = "direct"
+weight = 1
+enabled = true
+```
+
+### Пример 2: direct с interface + явными bind IP
+
+```toml
+[[upstreams]]
+type = "direct"
+interface = "eth0"
+bind_addresses = ["192.168.1.100", "192.168.1.101"]
+weight = 3
+enabled = true
+```
+
+### Пример 3: SOCKS5 upstream с аутентификацией
+
+```toml
+[[upstreams]]
+type = "socks5"
+address = "198.51.100.30:1080"
+username = "proxy-user"
+password = "proxy-pass"
+weight = 2
+enabled = true
+```
+
+### Пример 4: смешанные upstream с scopes
+
+```toml
+[[upstreams]]
+type = "direct"
+weight = 5
+enabled = true
+scopes = ""
+
+[[upstreams]]
+type = "socks5"
+address = "203.0.113.40:1080"
+username = "edge"
+password = "edgepass"
+weight = 3
+enabled = true
+scopes = "premium,me"
+```
+
+### Пример 5: профиль тюнинга под ME
+
+```toml
+[general]
+use_middle_proxy = true
+proxy_secret_path = "proxy-secret"
+middle_proxy_nat_probe = true
+stun_nat_probe_concurrency = 16
+middle_proxy_pool_size = 12
+me_keepalive_enabled = true
+me_keepalive_interval_secs = 20
+me_keepalive_jitter_secs = 4
+me_reconnect_max_concurrent_per_dc = 12
+me_reconnect_backoff_base_ms = 300
+me_reconnect_backoff_cap_ms = 10000
+me_reconnect_fast_retry_count = 10
+hardswap = true
+me_reinit_every_secs = 600
+me_hardswap_warmup_delay_min_ms = 500
+me_hardswap_warmup_delay_max_ms = 1200
+me_hardswap_warmup_extra_passes = 2
+me_hardswap_warmup_pass_backoff_base_ms = 400
+me_config_stable_snapshots = 3
+me_config_apply_cooldown_secs = 120
+proxy_secret_stable_snapshots = 3
+proxy_secret_rotate_runtime = true
+proxy_secret_len_max = 512
+update_every = 300
+me_pool_drain_ttl_secs = 120
+me_pool_min_fresh_ratio = 0.9
+me_reinit_drain_timeout_secs = 180
+
+[timeouts]
+me_one_retry = 8
+me_one_timeout_ms = 1200
+
+[network]
+stun_use = true
+stun_tcp_fallback = true
+stun_servers = [
+  "stun1.l.google.com:19302",
+  "stun2.l.google.com:19302"
+]
+http_ip_detect_urls = [
+  "https://api.ipify.org",
+  "https://ifconfig.me/ip"
+]
+```

proxy-secret

@@ -1 +0,0 @@
-ΔωϊΚ–xζ»H­l~,εΐ<CEB5>D0d]UJέλUA<55>M¦'!ΠFκ«nR«©ZD>Ο³F>y Zfa*ί<>®Ϊ‹ι¨

src/config/defaults.rs

@@ -3,6 +3,17 @@ use ipnetwork::IpNetwork;
 use serde::Deserialize;
 
 // Helper defaults kept private to the config module.
+const DEFAULT_NETWORK_IPV6: Option<bool> = Some(false);
+const DEFAULT_STUN_TCP_FALLBACK: bool = true;
+const DEFAULT_MIDDLE_PROXY_WARM_STANDBY: usize = 16;
+const DEFAULT_ME_RECONNECT_MAX_CONCURRENT_PER_DC: u32 = 8;
+const DEFAULT_ME_RECONNECT_FAST_RETRY_COUNT: u32 = 16;
+const DEFAULT_UPSTREAM_CONNECT_RETRY_ATTEMPTS: u32 = 3;
+const DEFAULT_UPSTREAM_UNHEALTHY_FAIL_THRESHOLD: u32 = 4;
+const DEFAULT_LISTEN_ADDR_IPV6: &str = "::";
+const DEFAULT_ACCESS_USER: &str = "default";
+const DEFAULT_ACCESS_SECRET: &str = "00000000000000000000000000000000";
+
 pub(crate) fn default_true() -> bool {
     true
 }
@@ -12,7 +23,7 @@ pub(crate) fn default_port() -> u16 {
 }
 
 pub(crate) fn default_tls_domain() -> String {
-    "www.google.com".to_string()
+    "petrovich.ru".to_string()
 }
 
 pub(crate) fn default_mask_port() -> u16 {
@@ -36,7 +47,7 @@ pub(crate) fn default_replay_window_secs() -> u64 {
 }
 
 pub(crate) fn default_handshake_timeout() -> u64 {
-    15
+    30
 }
 
 pub(crate) fn default_connect_timeout() -> u64 {
@@ -51,17 +62,21 @@ pub(crate) fn default_ack_timeout() -> u64 {
     300
 }
 pub(crate) fn default_me_one_retry() -> u8 {
-    3
+    12
 }
 
 pub(crate) fn default_me_one_timeout() -> u64 {
-    1500
+    1200
 }
 
 pub(crate) fn default_listen_addr() -> String {
     "0.0.0.0".to_string()
 }
 
+pub(crate) fn default_listen_addr_ipv4() -> Option<String> {
+    Some(default_listen_addr())
+}
+
 pub(crate) fn default_weight() -> u16 {
     1
 }
@@ -77,6 +92,14 @@ pub(crate) fn default_prefer_4() -> u8 {
     4
 }
 
+pub(crate) fn default_network_ipv6() -> Option<bool> {
+    DEFAULT_NETWORK_IPV6
+}
+
+pub(crate) fn default_stun_tcp_fallback() -> bool {
+    DEFAULT_STUN_TCP_FALLBACK
+}
+
 pub(crate) fn default_unknown_dc_log_path() -> Option<String> {
     Some("unknown-dc.txt".to_string())
 }
@@ -85,6 +108,26 @@ pub(crate) fn default_pool_size() -> usize {
     8
 }
 
+pub(crate) fn default_proxy_secret_path() -> Option<String> {
+    Some("proxy-secret".to_string())
+}
+
+pub(crate) fn default_middle_proxy_nat_stun() -> Option<String> {
+    None
+}
+
+pub(crate) fn default_middle_proxy_nat_stun_servers() -> Vec<String> {
+    Vec::new()
+}
+
+pub(crate) fn default_stun_nat_probe_concurrency() -> usize {
+    8
+}
+
+pub(crate) fn default_middle_proxy_warm_standby() -> usize {
+    DEFAULT_MIDDLE_PROXY_WARM_STANDBY
+}
+
 pub(crate) fn default_keepalive_interval() -> u64 {
     25
 }
@@ -109,6 +152,26 @@ pub(crate) fn default_reconnect_backoff_cap_ms() -> u64 {
     30_000
 }
 
+pub(crate) fn default_me_reconnect_max_concurrent_per_dc() -> u32 {
+    DEFAULT_ME_RECONNECT_MAX_CONCURRENT_PER_DC
+}
+
+pub(crate) fn default_me_reconnect_fast_retry_count() -> u32 {
+    DEFAULT_ME_RECONNECT_FAST_RETRY_COUNT
+}
+
+pub(crate) fn default_upstream_connect_retry_attempts() -> u32 {
+    DEFAULT_UPSTREAM_CONNECT_RETRY_ATTEMPTS
+}
+
+pub(crate) fn default_upstream_connect_retry_backoff_ms() -> u64 {
+    250
+}
+
+pub(crate) fn default_upstream_unhealthy_fail_threshold() -> u32 {
+    DEFAULT_UPSTREAM_UNHEALTHY_FAIL_THRESHOLD
+}
+
 pub(crate) fn default_crypto_pending_buffer() -> usize {
     256 * 1024
 }
@@ -121,6 +184,30 @@ pub(crate) fn default_desync_all_full() -> bool {
     false
 }
 
+pub(crate) fn default_me_route_backpressure_base_timeout_ms() -> u64 {
+    25
+}
+
+pub(crate) fn default_me_route_backpressure_high_timeout_ms() -> u64 {
+    120
+}
+
+pub(crate) fn default_me_route_backpressure_high_watermark_pct() -> u8 {
+    80
+}
+
+pub(crate) fn default_beobachten_minutes() -> u64 {
+    10
+}
+
+pub(crate) fn default_beobachten_flush_secs() -> u64 {
+    15
+}
+
+pub(crate) fn default_beobachten_file() -> String {
+    "cache/beobachten.txt".to_string()
+}
+
 pub(crate) fn default_tls_new_session_tickets() -> u8 {
     0
 }
@@ -179,7 +266,31 @@ pub(crate) fn default_proxy_config_reload_secs() -> u64 {
 }
 
 pub(crate) fn default_update_every_secs() -> u64 {
-    30 * 60
+    5 * 60
+}
+
+pub(crate) fn default_update_every() -> Option<u64> {
+    Some(default_update_every_secs())
+}
+
+pub(crate) fn default_me_reinit_every_secs() -> u64 {
+    15 * 60
+}
+
+pub(crate) fn default_me_hardswap_warmup_delay_min_ms() -> u64 {
+    1000
+}
+
+pub(crate) fn default_me_hardswap_warmup_delay_max_ms() -> u64 {
+    2000
+}
+
+pub(crate) fn default_me_hardswap_warmup_extra_passes() -> u8 {
+    3
+}
+
+pub(crate) fn default_me_hardswap_warmup_pass_backoff_base_ms() -> u64 {
+    500
 }
 
 pub(crate) fn default_me_config_stable_snapshots() -> u8 {
@@ -234,6 +345,21 @@ pub(crate) fn default_degradation_min_unavailable_dc_groups() -> u8 {
     2
 }
 
+pub(crate) fn default_listen_addr_ipv6() -> String {
+    DEFAULT_LISTEN_ADDR_IPV6.to_string()
+}
+
+pub(crate) fn default_listen_addr_ipv6_opt() -> Option<String> {
+    Some(default_listen_addr_ipv6())
+}
+
+pub(crate) fn default_access_users() -> HashMap<String, String> {
+    HashMap::from([(
+        DEFAULT_ACCESS_USER.to_string(),
+        DEFAULT_ACCESS_SECRET.to_string(),
+    )])
+}
+
 // Custom deserializer helpers
 
 #[derive(Deserialize)]

src/config/hot_reload.rs

@@ -16,6 +16,8 @@
 //! | `general` | `me_pool_drain_ttl_secs`      | Applied on next ME map update     |
 //! | `general` | `me_pool_min_fresh_ratio`     | Applied on next ME map update     |
 //! | `general` | `me_reinit_drain_timeout_secs`| Applied on next ME map update     |
+//! | `general` | `telemetry` / `me_*_policy`   | Applied immediately               |
+//! | `network` | `dns_overrides`               | Applied immediately               |
 //! | `access`  | All user/quota fields         | Effective immediately             |
 //!
 //! Fields that require re-binding sockets (`server.port`, `censorship.*`,
@@ -29,7 +31,7 @@ use notify::{EventKind, RecursiveMode, Watcher, recommended_watcher};
 use tokio::sync::{mpsc, watch};
 use tracing::{error, info, warn};
 
-use crate::config::LogLevel;
+use crate::config::{LogLevel, MeSocksKdfPolicy, MeTelemetryLevel};
 use super::load::ProxyConfig;
 
 // ── Hot fields ────────────────────────────────────────────────────────────────
@@ -39,6 +41,7 @@ use super::load::ProxyConfig;
 pub struct HotFields {
     pub log_level:               LogLevel,
     pub ad_tag:                  Option<String>,
+    pub dns_overrides:           Vec<String>,
     pub middle_proxy_pool_size:  usize,
     pub desync_all_full:         bool,
     pub update_every_secs:       u64,
@@ -50,6 +53,13 @@ pub struct HotFields {
     pub me_keepalive_interval_secs: u64,
     pub me_keepalive_jitter_secs:   u64,
     pub me_keepalive_payload_random: bool,
+    pub telemetry_core_enabled: bool,
+    pub telemetry_user_enabled: bool,
+    pub telemetry_me_level: MeTelemetryLevel,
+    pub me_socks_kdf_policy: MeSocksKdfPolicy,
+    pub me_route_backpressure_base_timeout_ms: u64,
+    pub me_route_backpressure_high_timeout_ms: u64,
+    pub me_route_backpressure_high_watermark_pct: u8,
     pub access:                  crate::config::AccessConfig,
 }
 
@@ -58,6 +68,7 @@ impl HotFields {
         Self {
             log_level:               cfg.general.log_level.clone(),
             ad_tag:                  cfg.general.ad_tag.clone(),
+            dns_overrides:           cfg.network.dns_overrides.clone(),
             middle_proxy_pool_size:  cfg.general.middle_proxy_pool_size,
             desync_all_full:         cfg.general.desync_all_full,
             update_every_secs:       cfg.general.effective_update_every_secs(),
@@ -69,6 +80,13 @@ impl HotFields {
             me_keepalive_interval_secs: cfg.general.me_keepalive_interval_secs,
             me_keepalive_jitter_secs:   cfg.general.me_keepalive_jitter_secs,
             me_keepalive_payload_random: cfg.general.me_keepalive_payload_random,
+            telemetry_core_enabled: cfg.general.telemetry.core_enabled,
+            telemetry_user_enabled: cfg.general.telemetry.user_enabled,
+            telemetry_me_level: cfg.general.telemetry.me_level,
+            me_socks_kdf_policy: cfg.general.me_socks_kdf_policy,
+            me_route_backpressure_base_timeout_ms: cfg.general.me_route_backpressure_base_timeout_ms,
+            me_route_backpressure_high_timeout_ms: cfg.general.me_route_backpressure_high_timeout_ms,
+            me_route_backpressure_high_watermark_pct: cfg.general.me_route_backpressure_high_watermark_pct,
             access:                  cfg.access.clone(),
         }
     }
@@ -96,6 +114,17 @@ fn warn_non_hot_changes(old: &ProxyConfig, new: &ProxyConfig) {
     if old.general.use_middle_proxy != new.general.use_middle_proxy {
         warn!("config reload: use_middle_proxy changed; restart required");
     }
+    if old.general.stun_nat_probe_concurrency != new.general.stun_nat_probe_concurrency {
+        warn!("config reload: general.stun_nat_probe_concurrency changed; restart required");
+    }
+    if old.general.upstream_connect_retry_attempts != new.general.upstream_connect_retry_attempts
+        || old.general.upstream_connect_retry_backoff_ms
+            != new.general.upstream_connect_retry_backoff_ms
+        || old.general.upstream_unhealthy_fail_threshold
+            != new.general.upstream_unhealthy_fail_threshold
+    {
+        warn!("config reload: general.upstream_* changed; restart required");
+    }
 }
 
 /// Resolve the public host for link generation — mirrors the logic in main.rs.
@@ -186,6 +215,13 @@ fn log_changes(
         );
     }
 
+    if old_hot.dns_overrides != new_hot.dns_overrides {
+        info!(
+            "config reload: network.dns_overrides updated ({} entries)",
+            new_hot.dns_overrides.len()
+        );
+    }
+
     if old_hot.middle_proxy_pool_size != new_hot.middle_proxy_pool_size {
         info!(
             "config reload: middle_proxy_pool_size: {} → {}",
@@ -249,6 +285,41 @@ fn log_changes(
         );
     }
 
+    if old_hot.telemetry_core_enabled != new_hot.telemetry_core_enabled
+        || old_hot.telemetry_user_enabled != new_hot.telemetry_user_enabled
+        || old_hot.telemetry_me_level != new_hot.telemetry_me_level
+    {
+        info!(
+            "config reload: telemetry: core_enabled={} user_enabled={} me_level={}",
+            new_hot.telemetry_core_enabled,
+            new_hot.telemetry_user_enabled,
+            new_hot.telemetry_me_level,
+        );
+    }
+
+    if old_hot.me_socks_kdf_policy != new_hot.me_socks_kdf_policy {
+        info!(
+            "config reload: me_socks_kdf_policy: {:?} → {:?}",
+            old_hot.me_socks_kdf_policy,
+            new_hot.me_socks_kdf_policy,
+        );
+    }
+
+    if old_hot.me_route_backpressure_base_timeout_ms
+        != new_hot.me_route_backpressure_base_timeout_ms
+        || old_hot.me_route_backpressure_high_timeout_ms
+            != new_hot.me_route_backpressure_high_timeout_ms
+        || old_hot.me_route_backpressure_high_watermark_pct
+            != new_hot.me_route_backpressure_high_watermark_pct
+    {
+        info!(
+            "config reload: me_route_backpressure: base={}ms high={}ms watermark={}%",
+            new_hot.me_route_backpressure_base_timeout_ms,
+            new_hot.me_route_backpressure_high_timeout_ms,
+            new_hot.me_route_backpressure_high_watermark_pct,
+        );
+    }
+
     if old_hot.access.users != new_hot.access.users {
         let mut added: Vec<&String> = new_hot.access.users.keys()
             .filter(|u| !old_hot.access.users.contains_key(*u))
@@ -351,6 +422,16 @@ fn reload_config(
         return;
     }
 
+    if old_hot.dns_overrides != new_hot.dns_overrides
+        && let Err(e) = crate::network::dns_overrides::install_entries(&new_hot.dns_overrides)
+    {
+        error!(
+            "config reload: invalid network.dns_overrides: {}; keeping old config",
+            e
+        );
+        return;
+    }
+
     warn_non_hot_changes(&old_cfg, &new_cfg);
     log_changes(&old_hot, &new_hot, &new_cfg, log_tx, detected_ip_v4, detected_ip_v6);
     config_tx.send(Arc::new(new_cfg)).ok();

src/config/load.rs

@@ -65,6 +65,33 @@ fn validate_network_cfg(net: &mut NetworkConfig) -> Result<()> {
     Ok(())
 }
 
+fn push_unique_nonempty(target: &mut Vec<String>, value: String) {
+    let trimmed = value.trim();
+    if trimmed.is_empty() {
+        return;
+    }
+    if !target.iter().any(|existing| existing == trimmed) {
+        target.push(trimmed.to_string());
+    }
+}
+
+fn is_valid_ad_tag(tag: &str) -> bool {
+    tag.len() == 32 && tag.chars().all(|ch| ch.is_ascii_hexdigit())
+}
+
+fn sanitize_ad_tag(ad_tag: &mut Option<String>) {
+    let Some(tag) = ad_tag.as_ref() else {
+        return;
+    };
+
+    if !is_valid_ad_tag(tag) {
+        warn!(
+            "Invalid general.ad_tag value, expected exactly 32 hex chars; ad_tag is disabled"
+        );
+        *ad_tag = None;
+    }
+}
+
 // ============= Main Config =============
 
 #[derive(Debug, Clone, Serialize, Deserialize, Default)]
@@ -116,8 +143,65 @@ impl ProxyConfig {
         let base_dir = path.as_ref().parent().unwrap_or(Path::new("."));
         let processed = preprocess_includes(&content, base_dir, 0)?;
 
-        let mut config: ProxyConfig =
+        let parsed_toml: toml::Value =
             toml::from_str(&processed).map_err(|e| ProxyError::Config(e.to_string()))?;
+        let general_table = parsed_toml
+            .get("general")
+            .and_then(|value| value.as_table());
+        let network_table = parsed_toml
+            .get("network")
+            .and_then(|value| value.as_table());
+        let update_every_is_explicit = general_table
+            .map(|table| table.contains_key("update_every"))
+            .unwrap_or(false);
+        let legacy_secret_is_explicit = general_table
+            .map(|table| table.contains_key("proxy_secret_auto_reload_secs"))
+            .unwrap_or(false);
+        let legacy_config_is_explicit = general_table
+            .map(|table| table.contains_key("proxy_config_auto_reload_secs"))
+            .unwrap_or(false);
+        let stun_servers_is_explicit = network_table
+            .map(|table| table.contains_key("stun_servers"))
+            .unwrap_or(false);
+
+        let mut config: ProxyConfig =
+            parsed_toml.try_into().map_err(|e| ProxyError::Config(e.to_string()))?;
+
+        if !update_every_is_explicit && (legacy_secret_is_explicit || legacy_config_is_explicit) {
+            config.general.update_every = None;
+        }
+
+        let legacy_nat_stun = config.general.middle_proxy_nat_stun.take();
+        let legacy_nat_stun_servers = std::mem::take(&mut config.general.middle_proxy_nat_stun_servers);
+        let legacy_nat_stun_used = legacy_nat_stun.is_some() || !legacy_nat_stun_servers.is_empty();
+        if stun_servers_is_explicit {
+            let mut explicit_stun_servers = Vec::new();
+            for stun in std::mem::take(&mut config.network.stun_servers) {
+                push_unique_nonempty(&mut explicit_stun_servers, stun);
+            }
+            config.network.stun_servers = explicit_stun_servers;
+
+            if legacy_nat_stun_used {
+                warn!("general.middle_proxy_nat_stun and general.middle_proxy_nat_stun_servers are ignored because network.stun_servers is explicitly set");
+            }
+        } else {
+            // Keep the default STUN pool unless network.stun_servers is explicitly overridden.
+            let mut unified_stun_servers = default_stun_servers();
+            if let Some(stun) = legacy_nat_stun {
+                push_unique_nonempty(&mut unified_stun_servers, stun);
+            }
+            for stun in legacy_nat_stun_servers {
+                push_unique_nonempty(&mut unified_stun_servers, stun);
+            }
+
+            config.network.stun_servers = unified_stun_servers;
+
+            if legacy_nat_stun_used {
+                warn!("general.middle_proxy_nat_stun and general.middle_proxy_nat_stun_servers are deprecated; use network.stun_servers");
+            }
+        }
+
+        sanitize_ad_tag(&mut config.general.ad_tag);
 
         if let Some(update_every) = config.general.update_every {
             if update_every == 0 {
@@ -147,6 +231,74 @@ impl ProxyConfig {
             }
         }
 
+        if config.general.stun_nat_probe_concurrency == 0 {
+            return Err(ProxyError::Config(
+                "general.stun_nat_probe_concurrency must be > 0".to_string(),
+            ));
+        }
+
+        if config.general.upstream_connect_retry_attempts == 0 {
+            return Err(ProxyError::Config(
+                "general.upstream_connect_retry_attempts must be > 0".to_string(),
+            ));
+        }
+
+        if config.general.upstream_unhealthy_fail_threshold == 0 {
+            return Err(ProxyError::Config(
+                "general.upstream_unhealthy_fail_threshold must be > 0".to_string(),
+            ));
+        }
+
+        if config.general.me_reinit_every_secs == 0 {
+            return Err(ProxyError::Config(
+                "general.me_reinit_every_secs must be > 0".to_string(),
+            ));
+        }
+
+        if config.general.beobachten_minutes == 0 {
+            return Err(ProxyError::Config(
+                "general.beobachten_minutes must be > 0".to_string(),
+            ));
+        }
+
+        if config.general.beobachten_flush_secs == 0 {
+            return Err(ProxyError::Config(
+                "general.beobachten_flush_secs must be > 0".to_string(),
+            ));
+        }
+
+        if config.general.beobachten_file.trim().is_empty() {
+            return Err(ProxyError::Config(
+                "general.beobachten_file cannot be empty".to_string(),
+            ));
+        }
+
+        if config.general.me_hardswap_warmup_delay_max_ms == 0 {
+            return Err(ProxyError::Config(
+                "general.me_hardswap_warmup_delay_max_ms must be > 0".to_string(),
+            ));
+        }
+
+        if config.general.me_hardswap_warmup_delay_min_ms
+            > config.general.me_hardswap_warmup_delay_max_ms
+        {
+            return Err(ProxyError::Config(
+                "general.me_hardswap_warmup_delay_min_ms must be <= general.me_hardswap_warmup_delay_max_ms".to_string(),
+            ));
+        }
+
+        if config.general.me_hardswap_warmup_extra_passes > 10 {
+            return Err(ProxyError::Config(
+                "general.me_hardswap_warmup_extra_passes must be within [0, 10]".to_string(),
+            ));
+        }
+
+        if config.general.me_hardswap_warmup_pass_backoff_base_ms == 0 {
+            return Err(ProxyError::Config(
+                "general.me_hardswap_warmup_pass_backoff_base_ms must be > 0".to_string(),
+            ));
+        }
+
         if config.general.me_config_stable_snapshots == 0 {
             return Err(ProxyError::Config(
                 "general.me_config_stable_snapshots must be > 0".to_string(),
@@ -171,6 +323,26 @@ impl ProxyConfig {
             ));
         }
 
+        if config.general.me_route_backpressure_base_timeout_ms == 0 {
+            return Err(ProxyError::Config(
+                "general.me_route_backpressure_base_timeout_ms must be > 0".to_string(),
+            ));
+        }
+
+        if config.general.me_route_backpressure_high_timeout_ms
+            < config.general.me_route_backpressure_base_timeout_ms
+        {
+            return Err(ProxyError::Config(
+                "general.me_route_backpressure_high_timeout_ms must be >= general.me_route_backpressure_base_timeout_ms".to_string(),
+            ));
+        }
+
+        if !(1..=100).contains(&config.general.me_route_backpressure_high_watermark_pct) {
+            return Err(ProxyError::Config(
+                "general.me_route_backpressure_high_watermark_pct must be within [1, 100]".to_string(),
+            ));
+        }
+
         if config.general.effective_me_pool_force_close_secs() > 0
             && config.general.effective_me_pool_force_close_secs()
                 < config.general.me_pool_drain_ttl_secs
@@ -259,6 +431,7 @@ impl ProxyConfig {
         }
 
         validate_network_cfg(&mut config.network)?;
+        crate::network::dns_overrides::validate_entries(&config.network.dns_overrides)?;
 
         if config.general.use_middle_proxy && config.network.ipv6 == Some(true) {
             warn!("IPv6 with Middle Proxy is experimental and may cause KDF address mismatch; consider disabling IPv6 or ME");
@@ -361,14 +534,18 @@ impl ProxyConfig {
 
         if let Some(tag) = &self.general.ad_tag {
             let zeros = "00000000000000000000000000000000";
+            if !is_valid_ad_tag(tag) {
+                return Err(ProxyError::Config(
+                    "general.ad_tag must be exactly 32 hex characters".to_string(),
+                ));
+            }
             if tag == zeros {
                 warn!("ad_tag is all zeros; register a valid proxy tag via @MTProxybot to enable sponsored channel");
             }
-            if tag.len() != 32 || tag.chars().any(|c| !c.is_ascii_hexdigit()) {
-                warn!("ad_tag is not a 32-char hex string; ensure you use value issued by @MTProxybot");
-            }
         }
 
+        crate::network::dns_overrides::validate_entries(&self.network.dns_overrides)?;
+
         Ok(())
     }
 }
@@ -377,6 +554,90 @@ impl ProxyConfig {
 mod tests {
     use super::*;
 
+    #[test]
+    fn serde_defaults_remain_unchanged_for_present_sections() {
+        let toml = r#"
+            [network]
+            [general]
+            [server]
+            [access]
+        "#;
+        let cfg: ProxyConfig = toml::from_str(toml).unwrap();
+
+        assert_eq!(cfg.network.ipv6, default_network_ipv6());
+        assert_eq!(cfg.network.stun_use, default_true());
+        assert_eq!(cfg.network.stun_tcp_fallback, default_stun_tcp_fallback());
+        assert_eq!(
+            cfg.general.middle_proxy_warm_standby,
+            default_middle_proxy_warm_standby()
+        );
+        assert_eq!(
+            cfg.general.me_reconnect_max_concurrent_per_dc,
+            default_me_reconnect_max_concurrent_per_dc()
+        );
+        assert_eq!(
+            cfg.general.me_reconnect_fast_retry_count,
+            default_me_reconnect_fast_retry_count()
+        );
+        assert_eq!(
+            cfg.general.upstream_connect_retry_attempts,
+            default_upstream_connect_retry_attempts()
+        );
+        assert_eq!(
+            cfg.general.upstream_connect_retry_backoff_ms,
+            default_upstream_connect_retry_backoff_ms()
+        );
+        assert_eq!(
+            cfg.general.upstream_unhealthy_fail_threshold,
+            default_upstream_unhealthy_fail_threshold()
+        );
+        assert_eq!(cfg.general.update_every, default_update_every());
+        assert_eq!(cfg.server.listen_addr_ipv4, default_listen_addr_ipv4());
+        assert_eq!(cfg.server.listen_addr_ipv6, default_listen_addr_ipv6_opt());
+        assert_eq!(cfg.access.users, default_access_users());
+    }
+
+    #[test]
+    fn impl_defaults_are_sourced_from_default_helpers() {
+        let network = NetworkConfig::default();
+        assert_eq!(network.ipv6, default_network_ipv6());
+        assert_eq!(network.stun_use, default_true());
+        assert_eq!(network.stun_tcp_fallback, default_stun_tcp_fallback());
+
+        let general = GeneralConfig::default();
+        assert_eq!(
+            general.middle_proxy_warm_standby,
+            default_middle_proxy_warm_standby()
+        );
+        assert_eq!(
+            general.me_reconnect_max_concurrent_per_dc,
+            default_me_reconnect_max_concurrent_per_dc()
+        );
+        assert_eq!(
+            general.me_reconnect_fast_retry_count,
+            default_me_reconnect_fast_retry_count()
+        );
+        assert_eq!(
+            general.upstream_connect_retry_attempts,
+            default_upstream_connect_retry_attempts()
+        );
+        assert_eq!(
+            general.upstream_connect_retry_backoff_ms,
+            default_upstream_connect_retry_backoff_ms()
+        );
+        assert_eq!(
+            general.upstream_unhealthy_fail_threshold,
+            default_upstream_unhealthy_fail_threshold()
+        );
+        assert_eq!(general.update_every, default_update_every());
+
+        let server = ServerConfig::default();
+        assert_eq!(server.listen_addr_ipv6, Some(default_listen_addr_ipv6()));
+
+        let access = AccessConfig::default();
+        assert_eq!(access.users, default_access_users());
+    }
+
     #[test]
     fn dc_overrides_allow_string_and_array() {
         let toml = r#"
@@ -480,6 +741,221 @@ mod tests {
         let _ = std::fs::remove_file(path);
     }
 
+    #[test]
+    fn stun_nat_probe_concurrency_zero_is_rejected() {
+        let toml = r#"
+            [general]
+            stun_nat_probe_concurrency = 0
+
+            [censorship]
+            tls_domain = "example.com"
+
+            [access.users]
+            user = "00000000000000000000000000000000"
+        "#;
+        let dir = std::env::temp_dir();
+        let path = dir.join("telemt_stun_nat_probe_concurrency_zero_test.toml");
+        std::fs::write(&path, toml).unwrap();
+        let err = ProxyConfig::load(&path).unwrap_err().to_string();
+        assert!(err.contains("general.stun_nat_probe_concurrency must be > 0"));
+        let _ = std::fs::remove_file(path);
+    }
+
+    #[test]
+    fn me_reinit_every_default_is_set() {
+        let toml = r#"
+            [censorship]
+            tls_domain = "example.com"
+
+            [access.users]
+            user = "00000000000000000000000000000000"
+        "#;
+        let dir = std::env::temp_dir();
+        let path = dir.join("telemt_me_reinit_every_default_test.toml");
+        std::fs::write(&path, toml).unwrap();
+        let cfg = ProxyConfig::load(&path).unwrap();
+        assert_eq!(
+            cfg.general.me_reinit_every_secs,
+            default_me_reinit_every_secs()
+        );
+        let _ = std::fs::remove_file(path);
+    }
+
+    #[test]
+    fn me_reinit_every_zero_is_rejected() {
+        let toml = r#"
+            [general]
+            me_reinit_every_secs = 0
+
+            [censorship]
+            tls_domain = "example.com"
+
+            [access.users]
+            user = "00000000000000000000000000000000"
+        "#;
+        let dir = std::env::temp_dir();
+        let path = dir.join("telemt_me_reinit_every_zero_test.toml");
+        std::fs::write(&path, toml).unwrap();
+        let err = ProxyConfig::load(&path).unwrap_err().to_string();
+        assert!(err.contains("general.me_reinit_every_secs must be > 0"));
+        let _ = std::fs::remove_file(path);
+    }
+
+    #[test]
+    fn upstream_connect_retry_attempts_zero_is_rejected() {
+        let toml = r#"
+            [general]
+            upstream_connect_retry_attempts = 0
+
+            [censorship]
+            tls_domain = "example.com"
+
+            [access.users]
+            user = "00000000000000000000000000000000"
+        "#;
+        let dir = std::env::temp_dir();
+        let path = dir.join("telemt_upstream_connect_retry_attempts_zero_test.toml");
+        std::fs::write(&path, toml).unwrap();
+        let err = ProxyConfig::load(&path).unwrap_err().to_string();
+        assert!(err.contains("general.upstream_connect_retry_attempts must be > 0"));
+        let _ = std::fs::remove_file(path);
+    }
+
+    #[test]
+    fn upstream_unhealthy_fail_threshold_zero_is_rejected() {
+        let toml = r#"
+            [general]
+            upstream_unhealthy_fail_threshold = 0
+
+            [censorship]
+            tls_domain = "example.com"
+
+            [access.users]
+            user = "00000000000000000000000000000000"
+        "#;
+        let dir = std::env::temp_dir();
+        let path = dir.join("telemt_upstream_unhealthy_fail_threshold_zero_test.toml");
+        std::fs::write(&path, toml).unwrap();
+        let err = ProxyConfig::load(&path).unwrap_err().to_string();
+        assert!(err.contains("general.upstream_unhealthy_fail_threshold must be > 0"));
+        let _ = std::fs::remove_file(path);
+    }
+
+    #[test]
+    fn me_hardswap_warmup_defaults_are_set() {
+        let toml = r#"
+            [censorship]
+            tls_domain = "example.com"
+
+            [access.users]
+            user = "00000000000000000000000000000000"
+        "#;
+        let dir = std::env::temp_dir();
+        let path = dir.join("telemt_me_hardswap_warmup_defaults_test.toml");
+        std::fs::write(&path, toml).unwrap();
+        let cfg = ProxyConfig::load(&path).unwrap();
+        assert_eq!(
+            cfg.general.me_hardswap_warmup_delay_min_ms,
+            default_me_hardswap_warmup_delay_min_ms()
+        );
+        assert_eq!(
+            cfg.general.me_hardswap_warmup_delay_max_ms,
+            default_me_hardswap_warmup_delay_max_ms()
+        );
+        assert_eq!(
+            cfg.general.me_hardswap_warmup_extra_passes,
+            default_me_hardswap_warmup_extra_passes()
+        );
+        assert_eq!(
+            cfg.general.me_hardswap_warmup_pass_backoff_base_ms,
+            default_me_hardswap_warmup_pass_backoff_base_ms()
+        );
+        let _ = std::fs::remove_file(path);
+    }
+
+    #[test]
+    fn me_hardswap_warmup_delay_range_is_validated() {
+        let toml = r#"
+            [general]
+            me_hardswap_warmup_delay_min_ms = 2001
+            me_hardswap_warmup_delay_max_ms = 2000
+
+            [censorship]
+            tls_domain = "example.com"
+
+            [access.users]
+            user = "00000000000000000000000000000000"
+        "#;
+        let dir = std::env::temp_dir();
+        let path = dir.join("telemt_me_hardswap_warmup_delay_range_test.toml");
+        std::fs::write(&path, toml).unwrap();
+        let err = ProxyConfig::load(&path).unwrap_err().to_string();
+        assert!(err.contains(
+            "general.me_hardswap_warmup_delay_min_ms must be <= general.me_hardswap_warmup_delay_max_ms"
+        ));
+        let _ = std::fs::remove_file(path);
+    }
+
+    #[test]
+    fn me_hardswap_warmup_delay_max_zero_is_rejected() {
+        let toml = r#"
+            [general]
+            me_hardswap_warmup_delay_max_ms = 0
+
+            [censorship]
+            tls_domain = "example.com"
+
+            [access.users]
+            user = "00000000000000000000000000000000"
+        "#;
+        let dir = std::env::temp_dir();
+        let path = dir.join("telemt_me_hardswap_warmup_delay_max_zero_test.toml");
+        std::fs::write(&path, toml).unwrap();
+        let err = ProxyConfig::load(&path).unwrap_err().to_string();
+        assert!(err.contains("general.me_hardswap_warmup_delay_max_ms must be > 0"));
+        let _ = std::fs::remove_file(path);
+    }
+
+    #[test]
+    fn me_hardswap_warmup_extra_passes_out_of_range_is_rejected() {
+        let toml = r#"
+            [general]
+            me_hardswap_warmup_extra_passes = 11
+
+            [censorship]
+            tls_domain = "example.com"
+
+            [access.users]
+            user = "00000000000000000000000000000000"
+        "#;
+        let dir = std::env::temp_dir();
+        let path = dir.join("telemt_me_hardswap_warmup_extra_passes_test.toml");
+        std::fs::write(&path, toml).unwrap();
+        let err = ProxyConfig::load(&path).unwrap_err().to_string();
+        assert!(err.contains("general.me_hardswap_warmup_extra_passes must be within [0, 10]"));
+        let _ = std::fs::remove_file(path);
+    }
+
+    #[test]
+    fn me_hardswap_warmup_pass_backoff_zero_is_rejected() {
+        let toml = r#"
+            [general]
+            me_hardswap_warmup_pass_backoff_base_ms = 0
+
+            [censorship]
+            tls_domain = "example.com"
+
+            [access.users]
+            user = "00000000000000000000000000000000"
+        "#;
+        let dir = std::env::temp_dir();
+        let path = dir.join("telemt_me_hardswap_warmup_backoff_zero_test.toml");
+        std::fs::write(&path, toml).unwrap();
+        let err = ProxyConfig::load(&path).unwrap_err().to_string();
+        assert!(err.contains("general.me_hardswap_warmup_pass_backoff_base_ms must be > 0"));
+        let _ = std::fs::remove_file(path);
+    }
+
     #[test]
     fn me_config_stable_snapshots_zero_is_rejected() {
         let toml = r#"
@@ -580,4 +1056,87 @@ mod tests {
         assert_eq!(cfg.general.me_reinit_drain_timeout_secs, 90);
         let _ = std::fs::remove_file(path);
     }
+
+    #[test]
+    fn invalid_ad_tag_is_disabled_during_load() {
+        let toml = r#"
+            [general]
+            ad_tag = "not_hex"
+
+            [censorship]
+            tls_domain = "example.com"
+
+            [access.users]
+            user = "00000000000000000000000000000000"
+        "#;
+        let dir = std::env::temp_dir();
+        let path = dir.join("telemt_invalid_ad_tag_test.toml");
+        std::fs::write(&path, toml).unwrap();
+        let cfg = ProxyConfig::load(&path).unwrap();
+        assert!(cfg.general.ad_tag.is_none());
+        let _ = std::fs::remove_file(path);
+    }
+
+    #[test]
+    fn valid_ad_tag_is_preserved_during_load() {
+        let toml = r#"
+            [general]
+            ad_tag = "00112233445566778899aabbccddeeff"
+
+            [censorship]
+            tls_domain = "example.com"
+
+            [access.users]
+            user = "00000000000000000000000000000000"
+        "#;
+        let dir = std::env::temp_dir();
+        let path = dir.join("telemt_valid_ad_tag_test.toml");
+        std::fs::write(&path, toml).unwrap();
+        let cfg = ProxyConfig::load(&path).unwrap();
+        assert_eq!(
+            cfg.general.ad_tag.as_deref(),
+            Some("00112233445566778899aabbccddeeff")
+        );
+        let _ = std::fs::remove_file(path);
+    }
+
+    #[test]
+    fn invalid_dns_override_is_rejected() {
+        let toml = r#"
+            [network]
+            dns_overrides = ["example.com:443:2001:db8::10"]
+
+            [censorship]
+            tls_domain = "example.com"
+
+            [access.users]
+            user = "00000000000000000000000000000000"
+        "#;
+        let dir = std::env::temp_dir();
+        let path = dir.join("telemt_invalid_dns_override_test.toml");
+        std::fs::write(&path, toml).unwrap();
+        let err = ProxyConfig::load(&path).unwrap_err().to_string();
+        assert!(err.contains("must be bracketed"));
+        let _ = std::fs::remove_file(path);
+    }
+
+    #[test]
+    fn valid_dns_override_is_accepted() {
+        let toml = r#"
+            [network]
+            dns_overrides = ["example.com:443:127.0.0.1", "example.net:443:[2001:db8::10]"]
+
+            [censorship]
+            tls_domain = "example.com"
+
+            [access.users]
+            user = "00000000000000000000000000000000"
+        "#;
+        let dir = std::env::temp_dir();
+        let path = dir.join("telemt_valid_dns_override_test.toml");
+        std::fs::write(&path, toml).unwrap();
+        let cfg = ProxyConfig::load(&path).unwrap();
+        assert_eq!(cfg.network.dns_overrides.len(), 2);
+        let _ = std::fs::remove_file(path);
+    }
 }

src/config/types.rs

@@ -59,6 +59,98 @@ impl std::fmt::Display for LogLevel {
     }
 }
 
+/// Middle-End telemetry verbosity level.
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize, Default)]
+#[serde(rename_all = "lowercase")]
+pub enum MeTelemetryLevel {
+    #[default]
+    Normal,
+    Silent,
+    Debug,
+}
+
+impl MeTelemetryLevel {
+    pub fn as_u8(self) -> u8 {
+        match self {
+            MeTelemetryLevel::Silent => 0,
+            MeTelemetryLevel::Normal => 1,
+            MeTelemetryLevel::Debug => 2,
+        }
+    }
+
+    pub fn from_u8(raw: u8) -> Self {
+        match raw {
+            0 => MeTelemetryLevel::Silent,
+            2 => MeTelemetryLevel::Debug,
+            _ => MeTelemetryLevel::Normal,
+        }
+    }
+
+    pub fn allows_normal(self) -> bool {
+        !matches!(self, MeTelemetryLevel::Silent)
+    }
+
+    pub fn allows_debug(self) -> bool {
+        matches!(self, MeTelemetryLevel::Debug)
+    }
+}
+
+impl std::fmt::Display for MeTelemetryLevel {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        match self {
+            MeTelemetryLevel::Silent => write!(f, "silent"),
+            MeTelemetryLevel::Normal => write!(f, "normal"),
+            MeTelemetryLevel::Debug => write!(f, "debug"),
+        }
+    }
+}
+
+/// Middle-End SOCKS KDF fallback policy.
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize, Default)]
+#[serde(rename_all = "lowercase")]
+pub enum MeSocksKdfPolicy {
+    #[default]
+    Strict,
+    Compat,
+}
+
+impl MeSocksKdfPolicy {
+    pub fn as_u8(self) -> u8 {
+        match self {
+            MeSocksKdfPolicy::Strict => 0,
+            MeSocksKdfPolicy::Compat => 1,
+        }
+    }
+
+    pub fn from_u8(raw: u8) -> Self {
+        match raw {
+            1 => MeSocksKdfPolicy::Compat,
+            _ => MeSocksKdfPolicy::Strict,
+        }
+    }
+}
+
+/// Telemetry controls for hot-path counters and ME diagnostics.
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+pub struct TelemetryConfig {
+    #[serde(default = "default_true")]
+    pub core_enabled: bool,
+    #[serde(default = "default_true")]
+    pub user_enabled: bool,
+    #[serde(default)]
+    pub me_level: MeTelemetryLevel,
+}
+
+impl Default for TelemetryConfig {
+    fn default() -> Self {
+        Self {
+            core_enabled: default_true(),
+            user_enabled: default_true(),
+            me_level: MeTelemetryLevel::Normal,
+        }
+    }
+}
+
 // ============= Sub-Configs =============
 
 #[derive(Debug, Clone, Serialize, Deserialize)]
@@ -76,7 +168,7 @@ impl Default for ProxyModes {
         Self {
             classic: false,
             secure: false,
-            tls: true,
+            tls: default_true(),
         }
     }
 }
@@ -87,7 +179,7 @@ pub struct NetworkConfig {
     pub ipv4: bool,
 
     /// None = auto-detect IPv6 availability.
-    #[serde(default)]
+    #[serde(default = "default_network_ipv6")]
     pub ipv6: Option<bool>,
 
     /// 4 or 6.
@@ -97,12 +189,17 @@ pub struct NetworkConfig {
     #[serde(default)]
     pub multipath: bool,
 
+    /// Global switch for STUN probing.
+    /// When false, STUN is fully disabled and only non-STUN detection remains.
+    #[serde(default = "default_true")]
+    pub stun_use: bool,
+
     /// STUN servers list for public IP discovery.
     #[serde(default = "default_stun_servers")]
     pub stun_servers: Vec<String>,
 
     /// Enable TCP STUN fallback when UDP is blocked.
-    #[serde(default)]
+    #[serde(default = "default_stun_tcp_fallback")]
     pub stun_tcp_fallback: bool,
 
     /// HTTP-based public IP detection endpoints (fallback after STUN).
@@ -112,19 +209,26 @@ pub struct NetworkConfig {
     /// Cache file path for detected public IP.
     #[serde(default = "default_cache_public_ip_path")]
     pub cache_public_ip_path: String,
+
+    /// Runtime DNS overrides in `host:port:ip` format.
+    /// IPv6 IP values must be bracketed: `[2001:db8::1]`.
+    #[serde(default)]
+    pub dns_overrides: Vec<String>,
 }
 
 impl Default for NetworkConfig {
     fn default() -> Self {
         Self {
-            ipv4: true,
-            ipv6: Some(false),
-            prefer: 4,
+            ipv4: default_true(),
+            ipv6: default_network_ipv6(),
+            prefer: default_prefer_4(),
             multipath: false,
+            stun_use: default_true(),
             stun_servers: default_stun_servers(),
-            stun_tcp_fallback: true,
+            stun_tcp_fallback: default_stun_tcp_fallback(),
             http_ip_detect_urls: default_http_ip_detect_urls(),
             cache_public_ip_path: default_cache_public_ip_path(),
+            dns_overrides: Vec::new(),
         }
     }
 }
@@ -140,7 +244,7 @@ pub struct GeneralConfig {
     #[serde(default = "default_true")]
     pub fast_mode: bool,
 
-    #[serde(default)]
+    #[serde(default = "default_true")]
     pub use_middle_proxy: bool,
 
     #[serde(default)]
@@ -148,7 +252,7 @@ pub struct GeneralConfig {
 
     /// Path to proxy-secret binary file (auto-downloaded if absent).
     /// Infrastructure secret from https://core.telegram.org/getProxySecret.
-    #[serde(default)]
+    #[serde(default = "default_proxy_secret_path")]
     pub proxy_secret_path: Option<String>,
 
     /// Public IP override for middle-proxy NAT environments.
@@ -157,23 +261,29 @@ pub struct GeneralConfig {
     pub middle_proxy_nat_ip: Option<IpAddr>,
 
     /// Enable STUN-based NAT probing to discover public IP:port for ME KDF.
-    #[serde(default)]
+    #[serde(default = "default_true")]
     pub middle_proxy_nat_probe: bool,
 
-    /// Optional STUN server address (host:port) for NAT probing.
-    #[serde(default)]
+    /// Deprecated legacy single STUN server for NAT probing.
+    /// Use `network.stun_servers` instead.
+    #[serde(default = "default_middle_proxy_nat_stun")]
     pub middle_proxy_nat_stun: Option<String>,
 
-    /// Optional list of STUN servers for NAT probing fallback.
-    #[serde(default)]
+    /// Deprecated legacy STUN list for NAT probing fallback.
+    /// Use `network.stun_servers` instead.
+    #[serde(default = "default_middle_proxy_nat_stun_servers")]
     pub middle_proxy_nat_stun_servers: Vec<String>,
 
+    /// Maximum number of concurrent STUN probes during NAT detection.
+    #[serde(default = "default_stun_nat_probe_concurrency")]
+    pub stun_nat_probe_concurrency: usize,
+
     /// Desired size of active Middle-Proxy writer pool.
     #[serde(default = "default_pool_size")]
     pub middle_proxy_pool_size: usize,
 
     /// Number of warm standby ME connections kept pre-initialized.
-    #[serde(default)]
+    #[serde(default = "default_middle_proxy_warm_standby")]
     pub middle_proxy_warm_standby: usize,
 
     /// Enable ME keepalive padding frames.
@@ -206,6 +316,22 @@ pub struct GeneralConfig {
     #[serde(default = "default_desync_all_full")]
     pub desync_all_full: bool,
 
+    /// Enable per-IP forensic observation buckets for scanners and handshake failures.
+    #[serde(default = "default_true")]
+    pub beobachten: bool,
+
+    /// Observation retention window in minutes for per-IP forensic buckets.
+    #[serde(default = "default_beobachten_minutes")]
+    pub beobachten_minutes: u64,
+
+    /// Snapshot flush interval in seconds for beob output file.
+    #[serde(default = "default_beobachten_flush_secs")]
+    pub beobachten_flush_secs: u64,
+
+    /// Snapshot file path for beob output.
+    #[serde(default = "default_beobachten_file")]
+    pub beobachten_file: String,
+
     /// Enable C-like hard-swap for ME pool generations.
     /// When true, Telemt prewarms a new generation and switches once full coverage is reached.
     #[serde(default = "default_hardswap")]
@@ -224,7 +350,7 @@ pub struct GeneralConfig {
     pub me_warmup_step_jitter_ms: u64,
 
     /// Max concurrent reconnect attempts per DC.
-    #[serde(default)]
+    #[serde(default = "default_me_reconnect_max_concurrent_per_dc")]
     pub me_reconnect_max_concurrent_per_dc: u32,
 
     /// Base backoff in ms for reconnect.
@@ -236,9 +362,21 @@ pub struct GeneralConfig {
     pub me_reconnect_backoff_cap_ms: u64,
 
     /// Fast retry attempts before backoff.
-    #[serde(default)]
+    #[serde(default = "default_me_reconnect_fast_retry_count")]
     pub me_reconnect_fast_retry_count: u32,
 
+    /// Connect attempts for the selected upstream before returning error/fallback.
+    #[serde(default = "default_upstream_connect_retry_attempts")]
+    pub upstream_connect_retry_attempts: u32,
+
+    /// Delay in milliseconds between upstream connect attempts.
+    #[serde(default = "default_upstream_connect_retry_backoff_ms")]
+    pub upstream_connect_retry_backoff_ms: u64,
+
+    /// Consecutive failed requests before upstream is marked unhealthy.
+    #[serde(default = "default_upstream_unhealthy_fail_threshold")]
+    pub upstream_unhealthy_fail_threshold: u32,
+
     /// Ignore STUN/interface IP mismatch (keep using Middle Proxy even if NAT detected).
     #[serde(default)]
     pub stun_iface_mismatch_ignore: bool,
@@ -254,6 +392,26 @@ pub struct GeneralConfig {
     #[serde(default)]
     pub disable_colors: bool,
 
+    /// Runtime telemetry controls for counters/metrics in hot paths.
+    #[serde(default)]
+    pub telemetry: TelemetryConfig,
+
+    /// SOCKS-bound KDF policy for Middle-End handshake.
+    #[serde(default)]
+    pub me_socks_kdf_policy: MeSocksKdfPolicy,
+
+    /// Base backpressure timeout in milliseconds for ME route channel send.
+    #[serde(default = "default_me_route_backpressure_base_timeout_ms")]
+    pub me_route_backpressure_base_timeout_ms: u64,
+
+    /// High backpressure timeout in milliseconds when queue occupancy is above watermark.
+    #[serde(default = "default_me_route_backpressure_high_timeout_ms")]
+    pub me_route_backpressure_high_timeout_ms: u64,
+
+    /// Queue occupancy percent threshold for high backpressure timeout.
+    #[serde(default = "default_me_route_backpressure_high_watermark_pct")]
+    pub me_route_backpressure_high_watermark_pct: u8,
+
     /// [general.links] — proxy link generation overrides.
     #[serde(default)]
     pub links: LinksConfig,
@@ -264,9 +422,29 @@ pub struct GeneralConfig {
 
     /// Unified ME updater interval in seconds for getProxyConfig/getProxyConfigV6/getProxySecret.
     /// When omitted, effective value falls back to legacy proxy_*_auto_reload_secs fields.
-    #[serde(default)]
+    #[serde(default = "default_update_every")]
     pub update_every: Option<u64>,
 
+    /// Periodic ME pool reinitialization interval in seconds.
+    #[serde(default = "default_me_reinit_every_secs")]
+    pub me_reinit_every_secs: u64,
+
+    /// Minimum delay in ms between hardswap warmup connect attempts.
+    #[serde(default = "default_me_hardswap_warmup_delay_min_ms")]
+    pub me_hardswap_warmup_delay_min_ms: u64,
+
+    /// Maximum delay in ms between hardswap warmup connect attempts.
+    #[serde(default = "default_me_hardswap_warmup_delay_max_ms")]
+    pub me_hardswap_warmup_delay_max_ms: u64,
+
+    /// Additional warmup passes in the same hardswap cycle after the base pass.
+    #[serde(default = "default_me_hardswap_warmup_extra_passes")]
+    pub me_hardswap_warmup_extra_passes: u8,
+
+    /// Base backoff in ms between hardswap warmup passes when floor is still incomplete.
+    #[serde(default = "default_me_hardswap_warmup_pass_backoff_base_ms")]
+    pub me_hardswap_warmup_pass_backoff_base_ms: u64,
+
     /// Number of identical getProxyConfig snapshots required before applying ME map updates.
     #[serde(default = "default_me_config_stable_snapshots")]
     pub me_config_stable_snapshots: u8,
@@ -334,38 +512,56 @@ impl Default for GeneralConfig {
         Self {
             modes: ProxyModes::default(),
             prefer_ipv6: false,
-            fast_mode: true,
-            use_middle_proxy: false,
+            fast_mode: default_true(),
+            use_middle_proxy: default_true(),
             ad_tag: None,
-            proxy_secret_path: None,
+            proxy_secret_path: default_proxy_secret_path(),
             middle_proxy_nat_ip: None,
-            middle_proxy_nat_probe: false,
-            middle_proxy_nat_stun: None,
-            middle_proxy_nat_stun_servers: Vec::new(),
+            middle_proxy_nat_probe: default_true(),
+            middle_proxy_nat_stun: default_middle_proxy_nat_stun(),
+            middle_proxy_nat_stun_servers: default_middle_proxy_nat_stun_servers(),
+            stun_nat_probe_concurrency: default_stun_nat_probe_concurrency(),
             middle_proxy_pool_size: default_pool_size(),
-            middle_proxy_warm_standby: 16,
-            me_keepalive_enabled: true,
+            middle_proxy_warm_standby: default_middle_proxy_warm_standby(),
+            me_keepalive_enabled: default_true(),
             me_keepalive_interval_secs: default_keepalive_interval(),
             me_keepalive_jitter_secs: default_keepalive_jitter(),
-            me_keepalive_payload_random: true,
-            me_warmup_stagger_enabled: true,
+            me_keepalive_payload_random: default_true(),
+            me_warmup_stagger_enabled: default_true(),
             me_warmup_step_delay_ms: default_warmup_step_delay_ms(),
             me_warmup_step_jitter_ms: default_warmup_step_jitter_ms(),
-            me_reconnect_max_concurrent_per_dc: 8,
+            me_reconnect_max_concurrent_per_dc: default_me_reconnect_max_concurrent_per_dc(),
             me_reconnect_backoff_base_ms: default_reconnect_backoff_base_ms(),
             me_reconnect_backoff_cap_ms: default_reconnect_backoff_cap_ms(),
-            me_reconnect_fast_retry_count: 8,
+            me_reconnect_fast_retry_count: default_me_reconnect_fast_retry_count(),
+            upstream_connect_retry_attempts: default_upstream_connect_retry_attempts(),
+            upstream_connect_retry_backoff_ms: default_upstream_connect_retry_backoff_ms(),
+            upstream_unhealthy_fail_threshold: default_upstream_unhealthy_fail_threshold(),
             stun_iface_mismatch_ignore: false,
             unknown_dc_log_path: default_unknown_dc_log_path(),
             log_level: LogLevel::Normal,
             disable_colors: false,
+            telemetry: TelemetryConfig::default(),
+            me_socks_kdf_policy: MeSocksKdfPolicy::Strict,
+            me_route_backpressure_base_timeout_ms: default_me_route_backpressure_base_timeout_ms(),
+            me_route_backpressure_high_timeout_ms: default_me_route_backpressure_high_timeout_ms(),
+            me_route_backpressure_high_watermark_pct: default_me_route_backpressure_high_watermark_pct(),
             links: LinksConfig::default(),
             crypto_pending_buffer: default_crypto_pending_buffer(),
             max_client_frame: default_max_client_frame(),
             desync_all_full: default_desync_all_full(),
+            beobachten: default_true(),
+            beobachten_minutes: default_beobachten_minutes(),
+            beobachten_flush_secs: default_beobachten_flush_secs(),
+            beobachten_file: default_beobachten_file(),
             hardswap: default_hardswap(),
             fast_mode_min_tls_record: default_fast_mode_min_tls_record(),
-            update_every: Some(default_update_every_secs()),
+            update_every: default_update_every(),
+            me_reinit_every_secs: default_me_reinit_every_secs(),
+            me_hardswap_warmup_delay_min_ms: default_me_hardswap_warmup_delay_min_ms(),
+            me_hardswap_warmup_delay_max_ms: default_me_hardswap_warmup_delay_max_ms(),
+            me_hardswap_warmup_extra_passes: default_me_hardswap_warmup_extra_passes(),
+            me_hardswap_warmup_pass_backoff_base_ms: default_me_hardswap_warmup_pass_backoff_base_ms(),
             me_config_stable_snapshots: default_me_config_stable_snapshots(),
             me_config_apply_cooldown_secs: default_me_config_apply_cooldown_secs(),
             proxy_secret_stable_snapshots: default_proxy_secret_stable_snapshots(),
@@ -378,7 +574,7 @@ impl Default for GeneralConfig {
             proxy_config_auto_reload_secs: default_proxy_config_reload_secs(),
             ntp_check: default_ntp_check(),
             ntp_servers: default_ntp_servers(),
-            auto_degradation_enabled: true,
+            auto_degradation_enabled: default_true(),
             degradation_min_unavailable_dc_groups: default_degradation_min_unavailable_dc_groups(),
         }
     }
@@ -392,6 +588,11 @@ impl GeneralConfig {
             .unwrap_or_else(|| self.proxy_secret_auto_reload_secs.min(self.proxy_config_auto_reload_secs))
     }
 
+    /// Resolve periodic zero-downtime reinit interval for ME writers.
+    pub fn effective_me_reinit_every_secs(&self) -> u64 {
+        self.me_reinit_every_secs
+    }
+
     /// Resolve force-close timeout for stale writers.
     /// `me_reinit_drain_timeout_secs` remains backward-compatible alias.
     pub fn effective_me_pool_force_close_secs(&self) -> u64 {
@@ -400,11 +601,11 @@ impl GeneralConfig {
 }
 
 /// `[general.links]` — proxy link generation settings.
-#[derive(Debug, Clone, Serialize, Deserialize, Default)]
+#[derive(Debug, Clone, Serialize, Deserialize)]
 pub struct LinksConfig {
     /// List of usernames whose tg:// links to display at startup.
     /// `"*"` = all users, `["alice", "bob"]` = specific users.
-    #[serde(default)]
+    #[serde(default = "default_links_show")]
     pub show: ShowLink,
 
     /// Public hostname/IP for tg:// link generation (overrides detected IP).
@@ -416,15 +617,25 @@ pub struct LinksConfig {
     pub public_port: Option<u16>,
 }
 
+impl Default for LinksConfig {
+    fn default() -> Self {
+        Self {
+            show: default_links_show(),
+            public_host: None,
+            public_port: None,
+        }
+    }
+}
+
 #[derive(Debug, Clone, Serialize, Deserialize)]
 pub struct ServerConfig {
     #[serde(default = "default_port")]
     pub port: u16,
 
-    #[serde(default)]
+    #[serde(default = "default_listen_addr_ipv4")]
     pub listen_addr_ipv4: Option<String>,
 
-    #[serde(default)]
+    #[serde(default = "default_listen_addr_ipv6_opt")]
     pub listen_addr_ipv6: Option<String>,
 
     #[serde(default)]
@@ -459,8 +670,8 @@ impl Default for ServerConfig {
     fn default() -> Self {
         Self {
             port: default_port(),
-            listen_addr_ipv4: Some(default_listen_addr()),
-            listen_addr_ipv6: Some("::".to_string()),
+            listen_addr_ipv4: default_listen_addr_ipv4(),
+            listen_addr_ipv6: default_listen_addr_ipv6_opt(),
             listen_unix_sock: None,
             listen_unix_sock_perm: None,
             listen_tcp: None,
@@ -533,7 +744,7 @@ pub struct AntiCensorshipConfig {
     pub fake_cert_len: usize,
 
     /// Enable TLS certificate emulation using cached real certificates.
-    #[serde(default)]
+    #[serde(default = "default_true")]
     pub tls_emulation: bool,
 
     /// Directory to store TLS front cache (on disk).
@@ -561,6 +772,12 @@ pub struct AntiCensorshipConfig {
     /// Enforce ALPN echo of client preference.
     #[serde(default = "default_alpn_enforce")]
     pub alpn_enforce: bool,
+
+    /// Send PROXY protocol header when connecting to mask_host.
+    /// 0 = disabled, 1 = v1 (text), 2 = v2 (binary).
+    /// Allows the backend to see the real client IP.
+    #[serde(default)]
+    pub mask_proxy_protocol: u8,
 }
 
 impl Default for AntiCensorshipConfig {
@@ -568,25 +785,26 @@ impl Default for AntiCensorshipConfig {
         Self {
             tls_domain: default_tls_domain(),
             tls_domains: Vec::new(),
-            mask: true,
+            mask: default_true(),
             mask_host: None,
             mask_port: default_mask_port(),
             mask_unix_sock: None,
             fake_cert_len: default_fake_cert_len(),
-            tls_emulation: false,
+            tls_emulation: true,
             tls_front_dir: default_tls_front_dir(),
             server_hello_delay_min_ms: default_server_hello_delay_min_ms(),
             server_hello_delay_max_ms: default_server_hello_delay_max_ms(),
             tls_new_session_tickets: default_tls_new_session_tickets(),
             tls_full_cert_ttl_secs: default_tls_full_cert_ttl_secs(),
             alpn_enforce: default_alpn_enforce(),
+            mask_proxy_protocol: 0,
         }
     }
 }
 
 #[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
 pub struct AccessConfig {
-    #[serde(default)]
+    #[serde(default = "default_access_users")]
     pub users: HashMap<String, String>,
 
     #[serde(default)]
@@ -613,13 +831,8 @@ pub struct AccessConfig {
 
 impl Default for AccessConfig {
     fn default() -> Self {
-        let mut users = HashMap::new();
-        users.insert(
-            "default".to_string(),
-            "00000000000000000000000000000000".to_string(),
-        );
         Self {
-            users,
+            users: default_access_users(),
             user_max_tcp_conns: HashMap::new(),
             user_expirations: HashMap::new(),
             user_data_quota: HashMap::new(),
@@ -701,7 +914,7 @@ pub struct ListenerConfig {
 /// In TOML, this can be:
 /// - `show_link = "*"`          — show links for all users
 /// - `show_link = ["a", "b"]`   — show links for specific users
-/// - omitted                    — show no links (default)
+/// - omitted                    — default depends on the owning config field
 #[derive(Debug, Clone, Default)]
 pub enum ShowLink {
     /// Don't show any links (default when omitted).
@@ -713,6 +926,10 @@ pub enum ShowLink {
     Specific(Vec<String>),
 }
 
+fn default_links_show() -> ShowLink {
+    ShowLink::All
+}
+
 impl ShowLink {
     /// Returns true if no links should be shown.
     pub fn is_empty(&self) -> bool {

src/main.rs

@@ -35,10 +35,13 @@ use crate::crypto::SecureRandom;
 use crate::ip_tracker::UserIpTracker;
 use crate::network::probe::{decide_network_capabilities, log_probe_result, run_probe};
 use crate::proxy::ClientHandler;
+use crate::stats::beobachten::BeobachtenStore;
+use crate::stats::telemetry::TelemetryPolicy;
 use crate::stats::{ReplayChecker, Stats};
 use crate::stream::BufferPool;
 use crate::transport::middle_proxy::{
     MePool, fetch_proxy_config, run_me_ping, MePingFamily, MePingSample, format_sample_line,
+    format_me_route,
 };
 use crate::transport::{ListenOptions, UpstreamManager, create_listener, find_listener_processes};
 use crate::tls_front::TlsFrontCache;
@@ -159,6 +162,15 @@ fn print_proxy_links(host: &str, port: u16, config: &ProxyConfig) {
     info!(target: "telemt::links", "------------------------");
 }
 
+async fn write_beobachten_snapshot(path: &str, payload: &str) -> std::io::Result<()> {
+    if let Some(parent) = std::path::Path::new(path).parent()
+        && !parent.as_os_str().is_empty()
+    {
+        tokio::fs::create_dir_all(parent).await?;
+    }
+    tokio::fs::write(path, payload).await
+}
+
 #[tokio::main]
 async fn main() -> std::result::Result<(), Box<dyn std::error::Error>> {
     let (config_path, cli_silent, cli_log_level) = parse_cli();
@@ -183,6 +195,11 @@ async fn main() -> std::result::Result<(), Box<dyn std::error::Error>> {
         std::process::exit(1);
     }
 
+    if let Err(e) = crate::network::dns_overrides::install_entries(&config.network.dns_overrides) {
+        eprintln!("[telemt] Invalid network.dns_overrides: {}", e);
+        std::process::exit(1);
+    }
+
     let has_rust_log = std::env::var("RUST_LOG").is_ok();
     let effective_log_level = if cli_silent {
         LogLevel::Silent
@@ -193,14 +210,14 @@ async fn main() -> std::result::Result<(), Box<dyn std::error::Error>> {
     };
 
     let (filter_layer, filter_handle) = reload::Layer::new(EnvFilter::new("info"));
-    
+
     // Configure color output based on config
     let fmt_layer = if config.general.disable_colors {
         fmt::Layer::default().with_ansi(false)
     } else {
         fmt::Layer::default().with_ansi(true)
     };
-    
+
     tracing_subscriber::registry()
         .with(filter_layer)
         .with(fmt_layer)
@@ -244,10 +261,149 @@ async fn main() -> std::result::Result<(), Box<dyn std::error::Error>> {
         warn!("Using default tls_domain. Consider setting a custom domain.");
     }
 
+    let upstream_manager = Arc::new(UpstreamManager::new(
+        config.upstreams.clone(),
+        config.general.upstream_connect_retry_attempts,
+        config.general.upstream_connect_retry_backoff_ms,
+        config.general.upstream_unhealthy_fail_threshold,
+    ));
+
+    let mut tls_domains = Vec::with_capacity(1 + config.censorship.tls_domains.len());
+    tls_domains.push(config.censorship.tls_domain.clone());
+    for d in &config.censorship.tls_domains {
+        if !tls_domains.contains(d) {
+            tls_domains.push(d.clone());
+        }
+    }
+
+    // Start TLS front fetching in background immediately, in parallel with STUN probing.
+    let tls_cache: Option<Arc<TlsFrontCache>> = if config.censorship.tls_emulation {
+        let cache = Arc::new(TlsFrontCache::new(
+            &tls_domains,
+            config.censorship.fake_cert_len,
+            &config.censorship.tls_front_dir,
+        ));
+        cache.load_from_disk().await;
+
+        let port = config.censorship.mask_port;
+        let proxy_protocol = config.censorship.mask_proxy_protocol;
+        let mask_host = config
+            .censorship
+            .mask_host
+            .clone()
+            .unwrap_or_else(|| config.censorship.tls_domain.clone());
+        let mask_unix_sock = config.censorship.mask_unix_sock.clone();
+        let fetch_timeout = Duration::from_secs(5);
+
+        let cache_initial = cache.clone();
+        let domains_initial = tls_domains.clone();
+        let host_initial = mask_host.clone();
+        let unix_sock_initial = mask_unix_sock.clone();
+        let upstream_initial = upstream_manager.clone();
+        tokio::spawn(async move {
+            let mut join = tokio::task::JoinSet::new();
+            for domain in domains_initial {
+                let cache_domain = cache_initial.clone();
+                let host_domain = host_initial.clone();
+                let unix_sock_domain = unix_sock_initial.clone();
+                let upstream_domain = upstream_initial.clone();
+                join.spawn(async move {
+                    match crate::tls_front::fetcher::fetch_real_tls(
+                        &host_domain,
+                        port,
+                        &domain,
+                        fetch_timeout,
+                        Some(upstream_domain),
+                        proxy_protocol,
+                        unix_sock_domain.as_deref(),
+                    )
+                    .await
+                    {
+                        Ok(res) => cache_domain.update_from_fetch(&domain, res).await,
+                        Err(e) => {
+                            warn!(domain = %domain, error = %e, "TLS emulation initial fetch failed")
+                        }
+                    }
+                });
+            }
+            while let Some(res) = join.join_next().await {
+                if let Err(e) = res {
+                    warn!(error = %e, "TLS emulation initial fetch task join failed");
+                }
+            }
+        });
+
+        let cache_timeout = cache.clone();
+        let domains_timeout = tls_domains.clone();
+        let fake_cert_len = config.censorship.fake_cert_len;
+        tokio::spawn(async move {
+            tokio::time::sleep(fetch_timeout).await;
+            for domain in domains_timeout {
+                let cached = cache_timeout.get(&domain).await;
+                if cached.domain == "default" {
+                    warn!(
+                        domain = %domain,
+                        timeout_secs = fetch_timeout.as_secs(),
+                        fake_cert_len,
+                        "TLS-front fetch not ready within timeout; using cache/default fake cert fallback"
+                    );
+                }
+            }
+        });
+
+        // Periodic refresh with jitter.
+        let cache_refresh = cache.clone();
+        let domains_refresh = tls_domains.clone();
+        let host_refresh = mask_host.clone();
+        let unix_sock_refresh = mask_unix_sock.clone();
+        let upstream_refresh = upstream_manager.clone();
+        tokio::spawn(async move {
+            loop {
+                let base_secs = rand::rng().random_range(4 * 3600..=6 * 3600);
+                let jitter_secs = rand::rng().random_range(0..=7200);
+                tokio::time::sleep(Duration::from_secs(base_secs + jitter_secs)).await;
+
+                let mut join = tokio::task::JoinSet::new();
+                for domain in domains_refresh.clone() {
+                    let cache_domain = cache_refresh.clone();
+                    let host_domain = host_refresh.clone();
+                    let unix_sock_domain = unix_sock_refresh.clone();
+                    let upstream_domain = upstream_refresh.clone();
+                    join.spawn(async move {
+                        match crate::tls_front::fetcher::fetch_real_tls(
+                            &host_domain,
+                            port,
+                            &domain,
+                            fetch_timeout,
+                            Some(upstream_domain),
+                            proxy_protocol,
+                            unix_sock_domain.as_deref(),
+                        )
+                        .await
+                        {
+                            Ok(res) => cache_domain.update_from_fetch(&domain, res).await,
+                            Err(e) => warn!(domain = %domain, error = %e, "TLS emulation refresh failed"),
+                        }
+                    });
+                }
+
+                while let Some(res) = join.join_next().await {
+                    if let Err(e) = res {
+                        warn!(error = %e, "TLS emulation refresh task join failed");
+                    }
+                }
+            }
+        });
+
+        Some(cache)
+    } else {
+        None
+    };
+
     let probe = run_probe(
         &config.network,
-        config.general.middle_proxy_nat_stun.clone(),
         config.general.middle_proxy_nat_probe,
+        config.general.stun_nat_probe_concurrency,
     )
     .await?;
     let decision = decide_network_capabilities(&config.network, &probe);
@@ -256,6 +412,8 @@ async fn main() -> std::result::Result<(), Box<dyn std::error::Error>> {
     let prefer_ipv6 = decision.prefer_ipv6();
     let mut use_middle_proxy = config.general.use_middle_proxy && (decision.ipv4_me || decision.ipv6_me);
     let stats = Arc::new(Stats::new());
+    stats.apply_telemetry_policy(TelemetryPolicy::from_config(&config.general.telemetry));
+    let beobachten = Arc::new(BeobachtenStore::new());
     let rng = Arc::new(SecureRandom::new());
 
     // IP Tracker initialization
@@ -265,6 +423,12 @@ async fn main() -> std::result::Result<(), Box<dyn std::error::Error>> {
     if !config.access.user_max_unique_ips.is_empty() {
         info!("IP limits configured for {} users", config.access.user_max_unique_ips.len());
     }
+    if !config.network.dns_overrides.is_empty() {
+        info!(
+            "Runtime DNS overrides configured: {} entries",
+            config.network.dns_overrides.len()
+        );
+    }
 
     // Connection concurrency limit
     let max_connections = Arc::new(Semaphore::new(10_000));
@@ -279,14 +443,17 @@ async fn main() -> std::result::Result<(), Box<dyn std::error::Error>> {
     // =====================================================================
     let me_pool: Option<Arc<MePool>> = if use_middle_proxy {
         info!("=== Middle Proxy Mode ===");
+        let me_nat_probe = config.general.middle_proxy_nat_probe && config.network.stun_use;
+        if config.general.middle_proxy_nat_probe && !config.network.stun_use {
+            info!("Middle-proxy STUN probing disabled by network.stun_use=false");
+        }
 
         // ad_tag (proxy_tag) for advertising
-        let proxy_tag = config.general.ad_tag.as_ref().map(|tag| {
-            hex::decode(tag).unwrap_or_else(|_| {
-                warn!("Invalid ad_tag hex, middle proxy ad_tag will be empty");
-                Vec::new()
-            })
-        });
+        let proxy_tag = config
+            .general
+            .ad_tag
+            .as_ref()
+            .map(|tag| hex::decode(tag).expect("general.ad_tag must be validated before startup"));
 
         // =============================================================
         // CRITICAL: Download Telegram proxy-secret (NOT user secret!)
@@ -346,9 +513,10 @@ async fn main() -> std::result::Result<(), Box<dyn std::error::Error>> {
                     proxy_tag,
                     proxy_secret,
                     config.general.middle_proxy_nat_ip,
-                    config.general.middle_proxy_nat_probe,
-                    config.general.middle_proxy_nat_stun.clone(),
-                    config.general.middle_proxy_nat_stun_servers.clone(),
+                    me_nat_probe,
+                    None,
+                    config.network.stun_servers.clone(),
+                    config.general.stun_nat_probe_concurrency,
                     probe.detected_ipv6,
                     config.timeouts.me_one_retry,
                     config.timeouts.me_one_timeout_ms,
@@ -356,6 +524,7 @@ async fn main() -> std::result::Result<(), Box<dyn std::error::Error>> {
                     cfg_v6.map.clone(),
                     cfg_v4.default_dc.or(cfg_v6.default_dc),
                     decision.clone(),
+                    Some(upstream_manager.clone()),
                     rng.clone(),
                     stats.clone(),
                     config.general.me_keepalive_enabled,
@@ -373,41 +542,44 @@ async fn main() -> std::result::Result<(), Box<dyn std::error::Error>> {
                     config.general.me_pool_drain_ttl_secs,
                     config.general.effective_me_pool_force_close_secs(),
                     config.general.me_pool_min_fresh_ratio,
+                    config.general.me_hardswap_warmup_delay_min_ms,
+                    config.general.me_hardswap_warmup_delay_max_ms,
+                    config.general.me_hardswap_warmup_extra_passes,
+                    config.general.me_hardswap_warmup_pass_backoff_base_ms,
+                    config.general.me_socks_kdf_policy,
+                    config.general.me_route_backpressure_base_timeout_ms,
+                    config.general.me_route_backpressure_high_timeout_ms,
+                    config.general.me_route_backpressure_high_watermark_pct,
                 );
 
                 let pool_size = config.general.middle_proxy_pool_size.max(1);
-                match pool.init(pool_size, &rng).await {
-                    Ok(()) => {
-                        info!("Middle-End pool initialized successfully");
+                loop {
+                    match pool.init(pool_size, &rng).await {
+                        Ok(()) => {
+                            info!("Middle-End pool initialized successfully");
 
-                        // Phase 4: Start health monitor
-                        let pool_clone = pool.clone();
-                        let rng_clone = rng.clone();
-                        let min_conns = pool_size;
-                        tokio::spawn(async move {
-                            crate::transport::middle_proxy::me_health_monitor(
-                                pool_clone, rng_clone, min_conns,
-                            )
-                            .await;
-                        });
+                            // Phase 4: Start health monitor
+                            let pool_clone = pool.clone();
+                            let rng_clone = rng.clone();
+                            let min_conns = pool_size;
+                            tokio::spawn(async move {
+                                crate::transport::middle_proxy::me_health_monitor(
+                                    pool_clone, rng_clone, min_conns,
+                                )
+                                .await;
+                            });
 
-                        // Periodic ME connection rotation
-                        let pool_clone_rot = pool.clone();
-                        let rng_clone_rot = rng.clone();
-                        tokio::spawn(async move {
-                            crate::transport::middle_proxy::me_rotation_task(
-                                pool_clone_rot,
-                                rng_clone_rot,
-                                std::time::Duration::from_secs(1800),
-                            )
-                            .await;
-                        });
-
-                        Some(pool)
-                    }
-                    Err(e) => {
-                        error!(error = %e, "Failed to initialize ME pool. Falling back to direct mode.");
-                        None
+                            break Some(pool);
+                        }
+                        Err(e) => {
+                            warn!(
+                                error = %e,
+                                retry_in_secs = 2,
+                                "ME pool is not ready yet; retrying startup initialization"
+                            );
+                            pool.reset_stun_state();
+                            tokio::time::sleep(Duration::from_secs(2)).await;
+                        }
                     }
                 }
             }
@@ -439,77 +611,8 @@ async fn main() -> std::result::Result<(), Box<dyn std::error::Error>> {
         Duration::from_secs(config.access.replay_window_secs),
     ));
 
-    let upstream_manager = Arc::new(UpstreamManager::new(config.upstreams.clone()));
     let buffer_pool = Arc::new(BufferPool::with_config(16 * 1024, 4096));
 
-    // TLS front cache (optional emulation)
-    let mut tls_domains = Vec::with_capacity(1 + config.censorship.tls_domains.len());
-    tls_domains.push(config.censorship.tls_domain.clone());
-    for d in &config.censorship.tls_domains {
-        if !tls_domains.contains(d) {
-            tls_domains.push(d.clone());
-        }
-    }
-
-    let tls_cache: Option<Arc<TlsFrontCache>> = if config.censorship.tls_emulation {
-        let cache = Arc::new(TlsFrontCache::new(
-            &tls_domains,
-            config.censorship.fake_cert_len,
-            &config.censorship.tls_front_dir,
-        ));
-
-        cache.load_from_disk().await;
-
-        let port = config.censorship.mask_port;
-        let mask_host = config.censorship.mask_host.clone()
-            .unwrap_or_else(|| config.censorship.tls_domain.clone());
-        // Initial synchronous fetch to warm cache before serving clients.
-        for domain in tls_domains.clone() {
-            match crate::tls_front::fetcher::fetch_real_tls(
-                &mask_host,
-                port,
-                &domain,
-                Duration::from_secs(5),
-                Some(upstream_manager.clone()),
-            )
-            .await
-            {
-                Ok(res) => cache.update_from_fetch(&domain, res).await,
-                Err(e) => warn!(domain = %domain, error = %e, "TLS emulation fetch failed"),
-            }
-        }
-
-        // Periodic refresh with jitter.
-        let cache_clone = cache.clone();
-        let domains = tls_domains.clone();
-        let upstream_for_task = upstream_manager.clone();
-        tokio::spawn(async move {
-            loop {
-                let base_secs = rand::rng().random_range(4 * 3600..=6 * 3600);
-                let jitter_secs = rand::rng().random_range(0..=7200);
-                tokio::time::sleep(Duration::from_secs(base_secs + jitter_secs)).await;
-                for domain in &domains {
-                    match crate::tls_front::fetcher::fetch_real_tls(
-                        &mask_host,
-                        port,
-                        domain,
-                        Duration::from_secs(5),
-                        Some(upstream_for_task.clone()),
-                    )
-                    .await
-                    {
-                        Ok(res) => cache_clone.update_from_fetch(domain, res).await,
-                        Err(e) => warn!(domain = %domain, error = %e, "TLS emulation refresh failed"),
-                    }
-                }
-            }
-        });
-
-        Some(cache)
-    } else {
-        None
-    };
-
     // Middle-End ping before DC connectivity
     if let Some(ref pool) = me_pool {
         let me_results = run_me_ping(pool, &rng).await;
@@ -533,7 +636,15 @@ async fn main() -> std::result::Result<(), Box<dyn std::error::Error>> {
         } else {
             info!("  No ME connectivity");
         }
-        info!("  via direct");
+        let me_route = format_me_route(
+            &config.upstreams,
+            &me_results,
+            prefer_ipv6,
+            v4_ok,
+            v6_ok,
+        )
+        .await;
+        info!("  via {}", me_route);
         info!("============================================================");
 
         use std::collections::BTreeMap;
@@ -603,9 +714,9 @@ async fn main() -> std::result::Result<(), Box<dyn std::error::Error>> {
 				info!("  IPv4 in use / IPv6 is fallback");
 			}
 		} else if v6_works && !v4_works {
-			info!("  IPv6 only / IPv4 unavailable)");
+			info!("  IPv6 only / IPv4 unavailable");
 		} else if v4_works && !v6_works {
-			info!("  IPv4 only / IPv6 unavailable)");
+			info!("  IPv4 only / IPv6 unavailable");
 		} else if !v6_works && !v4_works {
 			info!("  No DC connectivity");
 		}
@@ -674,14 +785,8 @@ async fn main() -> std::result::Result<(), Box<dyn std::error::Error>> {
         rc_clone.run_periodic_cleanup().await;
     });
 
-    let detected_ip_v4: Option<std::net::IpAddr> = probe
-        .reflected_ipv4
-        .map(|s| s.ip())
-        .or_else(|| probe.detected_ipv4.map(std::net::IpAddr::V4));
-    let detected_ip_v6: Option<std::net::IpAddr> = probe
-        .reflected_ipv6
-        .map(|s| s.ip())
-        .or_else(|| probe.detected_ipv6.map(std::net::IpAddr::V6));
+    let detected_ip_v4: Option<std::net::IpAddr> = probe.detected_ipv4.map(std::net::IpAddr::V4);
+    let detected_ip_v6: Option<std::net::IpAddr> = probe.detected_ipv6.map(std::net::IpAddr::V6);
     debug!(
         "Detected IPs: v4={:?} v6={:?}",
         detected_ip_v4, detected_ip_v6
@@ -700,6 +805,47 @@ async fn main() -> std::result::Result<(), Box<dyn std::error::Error>> {
         detected_ip_v6,
     );
 
+    let stats_policy = stats.clone();
+    let mut config_rx_policy = config_rx.clone();
+    let me_pool_policy = me_pool.clone();
+    tokio::spawn(async move {
+        loop {
+            if config_rx_policy.changed().await.is_err() {
+                break;
+            }
+            let cfg = config_rx_policy.borrow_and_update().clone();
+            stats_policy.apply_telemetry_policy(TelemetryPolicy::from_config(&cfg.general.telemetry));
+            if let Some(pool) = &me_pool_policy {
+                pool.update_runtime_transport_policy(
+                    cfg.general.me_socks_kdf_policy,
+                    cfg.general.me_route_backpressure_base_timeout_ms,
+                    cfg.general.me_route_backpressure_high_timeout_ms,
+                    cfg.general.me_route_backpressure_high_watermark_pct,
+                );
+            }
+        }
+    });
+
+    let beobachten_writer = beobachten.clone();
+    let config_rx_beobachten = config_rx.clone();
+    tokio::spawn(async move {
+        loop {
+            let cfg = config_rx_beobachten.borrow().clone();
+            let sleep_secs = cfg.general.beobachten_flush_secs.max(1);
+
+            if cfg.general.beobachten {
+                let ttl = Duration::from_secs(cfg.general.beobachten_minutes.saturating_mul(60));
+                let path = cfg.general.beobachten_file.clone();
+                let snapshot = beobachten_writer.snapshot_text(ttl);
+                if let Err(e) = write_beobachten_snapshot(&path, &snapshot).await {
+                    warn!(error = %e, path = %path, "Failed to flush beobachten snapshot");
+                }
+            }
+
+            tokio::time::sleep(Duration::from_secs(sleep_secs)).await;
+        }
+    });
+
     if let Some(ref pool) = me_pool {
         let pool_clone = pool.clone();
         let rng_clone = rng.clone();
@@ -712,6 +858,18 @@ async fn main() -> std::result::Result<(), Box<dyn std::error::Error>> {
             )
             .await;
         });
+
+        let pool_clone_rot = pool.clone();
+        let rng_clone_rot = rng.clone();
+        let config_rx_clone_rot = config_rx.clone();
+        tokio::spawn(async move {
+            crate::transport::middle_proxy::me_rotation_task(
+                pool_clone_rot,
+                rng_clone_rot,
+                config_rx_clone_rot,
+            )
+            .await;
+        });
     }
 
     let mut listeners = Vec::new();
@@ -856,6 +1014,7 @@ async fn main() -> std::result::Result<(), Box<dyn std::error::Error>> {
         let me_pool = me_pool.clone();
         let tls_cache = tls_cache.clone();
         let ip_tracker = ip_tracker.clone();
+        let beobachten = beobachten.clone();
         let max_connections_unix = max_connections.clone();
 
         tokio::spawn(async move {
@@ -883,6 +1042,7 @@ async fn main() -> std::result::Result<(), Box<dyn std::error::Error>> {
                         let me_pool = me_pool.clone();
                         let tls_cache = tls_cache.clone();
                         let ip_tracker = ip_tracker.clone();
+                        let beobachten = beobachten.clone();
                         let proxy_protocol_enabled = config.server.proxy_protocol;
 
                         tokio::spawn(async move {
@@ -890,7 +1050,7 @@ async fn main() -> std::result::Result<(), Box<dyn std::error::Error>> {
                             if let Err(e) = crate::proxy::client::handle_client_stream(
                                 stream, fake_peer, config, stats,
                                 upstream_manager, replay_checker, buffer_pool, rng,
-                                me_pool, tls_cache, ip_tracker, proxy_protocol_enabled,
+                                me_pool, tls_cache, ip_tracker, beobachten, proxy_protocol_enabled,
                             ).await {
                                 debug!(error = %e, "Unix socket connection error");
                             }
@@ -938,9 +1098,20 @@ async fn main() -> std::result::Result<(), Box<dyn std::error::Error>> {
 
     if let Some(port) = config.server.metrics_port {
         let stats = stats.clone();
+        let beobachten = beobachten.clone();
+        let config_rx_metrics = config_rx.clone();
+        let ip_tracker_metrics = ip_tracker.clone();
         let whitelist = config.server.metrics_whitelist.clone();
         tokio::spawn(async move {
-            metrics::serve(port, stats, whitelist).await;
+            metrics::serve(
+                port,
+                stats,
+                beobachten,
+                ip_tracker_metrics,
+                config_rx_metrics,
+                whitelist,
+            )
+            .await;
         });
     }
 
@@ -954,6 +1125,7 @@ async fn main() -> std::result::Result<(), Box<dyn std::error::Error>> {
         let me_pool = me_pool.clone();
         let tls_cache = tls_cache.clone();
         let ip_tracker = ip_tracker.clone();
+        let beobachten = beobachten.clone();
         let max_connections_tcp = max_connections.clone();
 
         tokio::spawn(async move {
@@ -976,6 +1148,7 @@ async fn main() -> std::result::Result<(), Box<dyn std::error::Error>> {
                         let me_pool = me_pool.clone();
                         let tls_cache = tls_cache.clone();
                         let ip_tracker = ip_tracker.clone();
+                        let beobachten = beobachten.clone();
                         let proxy_protocol_enabled = listener_proxy_protocol;
 
                         tokio::spawn(async move {
@@ -992,6 +1165,7 @@ async fn main() -> std::result::Result<(), Box<dyn std::error::Error>> {
                                 me_pool,
                                 tls_cache,
                                 ip_tracker,
+                                beobachten,
                                 proxy_protocol_enabled,
                             )
                             .run()

src/metrics.rs

@@ -1,6 +1,8 @@
 use std::convert::Infallible;
+use std::collections::{BTreeSet, HashMap};
 use std::net::SocketAddr;
 use std::sync::Arc;
+use std::time::Duration;
 
 use http_body_util::Full;
 use hyper::body::Bytes;
@@ -11,9 +13,19 @@ use ipnetwork::IpNetwork;
 use tokio::net::TcpListener;
 use tracing::{info, warn, debug};
 
+use crate::config::ProxyConfig;
+use crate::ip_tracker::UserIpTracker;
+use crate::stats::beobachten::BeobachtenStore;
 use crate::stats::Stats;
 
-pub async fn serve(port: u16, stats: Arc<Stats>, whitelist: Vec<IpNetwork>) {
+pub async fn serve(
+    port: u16,
+    stats: Arc<Stats>,
+    beobachten: Arc<BeobachtenStore>,
+    ip_tracker: Arc<UserIpTracker>,
+    config_rx: tokio::sync::watch::Receiver<Arc<ProxyConfig>>,
+    whitelist: Vec<IpNetwork>,
+) {
     let addr = SocketAddr::from(([0, 0, 0, 0], port));
     let listener = match TcpListener::bind(addr).await {
         Ok(l) => l,
@@ -22,7 +34,7 @@ pub async fn serve(port: u16, stats: Arc<Stats>, whitelist: Vec<IpNetwork>) {
             return;
         }
     };
-    info!("Metrics endpoint: http://{}/metrics", addr);
+    info!("Metrics endpoint: http://{}/metrics and /beobachten", addr);
 
     loop {
         let (stream, peer) = match listener.accept().await {
@@ -39,10 +51,16 @@ pub async fn serve(port: u16, stats: Arc<Stats>, whitelist: Vec<IpNetwork>) {
         }
 
         let stats = stats.clone();
+        let beobachten = beobachten.clone();
+        let ip_tracker = ip_tracker.clone();
+        let config_rx_conn = config_rx.clone();
         tokio::spawn(async move {
             let svc = service_fn(move |req| {
                 let stats = stats.clone();
-                async move { handle(req, &stats) }
+                let beobachten = beobachten.clone();
+                let ip_tracker = ip_tracker.clone();
+                let config = config_rx_conn.borrow().clone();
+                async move { handle(req, &stats, &beobachten, &ip_tracker, &config).await }
             });
             if let Err(e) = http1::Builder::new()
                 .serve_connection(hyper_util::rt::TokioIo::new(stream), svc)
@@ -54,141 +72,440 @@ pub async fn serve(port: u16, stats: Arc<Stats>, whitelist: Vec<IpNetwork>) {
     }
 }
 
-fn handle<B>(req: Request<B>, stats: &Stats) -> Result<Response<Full<Bytes>>, Infallible> {
-    if req.uri().path() != "/metrics" {
+async fn handle<B>(
+    req: Request<B>,
+    stats: &Stats,
+    beobachten: &BeobachtenStore,
+    ip_tracker: &UserIpTracker,
+    config: &ProxyConfig,
+) -> Result<Response<Full<Bytes>>, Infallible> {
+    if req.uri().path() == "/metrics" {
+        let body = render_metrics(stats, config, ip_tracker).await;
         let resp = Response::builder()
-            .status(StatusCode::NOT_FOUND)
-            .body(Full::new(Bytes::from("Not Found\n")))
+            .status(StatusCode::OK)
+            .header("content-type", "text/plain; version=0.0.4; charset=utf-8")
+            .body(Full::new(Bytes::from(body)))
+            .unwrap();
+        return Ok(resp);
+    }
+
+    if req.uri().path() == "/beobachten" {
+        let body = render_beobachten(beobachten, config);
+        let resp = Response::builder()
+            .status(StatusCode::OK)
+            .header("content-type", "text/plain; charset=utf-8")
+            .body(Full::new(Bytes::from(body)))
             .unwrap();
         return Ok(resp);
     }
 
-    let body = render_metrics(stats);
     let resp = Response::builder()
-        .status(StatusCode::OK)
-        .header("content-type", "text/plain; version=0.0.4; charset=utf-8")
-        .body(Full::new(Bytes::from(body)))
+        .status(StatusCode::NOT_FOUND)
+        .body(Full::new(Bytes::from("Not Found\n")))
         .unwrap();
     Ok(resp)
 }
 
-fn render_metrics(stats: &Stats) -> String {
+fn render_beobachten(beobachten: &BeobachtenStore, config: &ProxyConfig) -> String {
+    if !config.general.beobachten {
+        return "beobachten disabled\n".to_string();
+    }
+
+    let ttl = Duration::from_secs(config.general.beobachten_minutes.saturating_mul(60));
+    beobachten.snapshot_text(ttl)
+}
+
+async fn render_metrics(stats: &Stats, config: &ProxyConfig, ip_tracker: &UserIpTracker) -> String {
     use std::fmt::Write;
     let mut out = String::with_capacity(4096);
+    let telemetry = stats.telemetry_policy();
+    let core_enabled = telemetry.core_enabled;
+    let user_enabled = telemetry.user_enabled;
+    let me_allows_normal = telemetry.me_level.allows_normal();
+    let me_allows_debug = telemetry.me_level.allows_debug();
 
     let _ = writeln!(out, "# HELP telemt_uptime_seconds Proxy uptime");
     let _ = writeln!(out, "# TYPE telemt_uptime_seconds gauge");
     let _ = writeln!(out, "telemt_uptime_seconds {:.1}", stats.uptime_secs());
 
+    let _ = writeln!(out, "# HELP telemt_telemetry_core_enabled Runtime core telemetry switch");
+    let _ = writeln!(out, "# TYPE telemt_telemetry_core_enabled gauge");
+    let _ = writeln!(
+        out,
+        "telemt_telemetry_core_enabled {}",
+        if core_enabled { 1 } else { 0 }
+    );
+
+    let _ = writeln!(out, "# HELP telemt_telemetry_user_enabled Runtime per-user telemetry switch");
+    let _ = writeln!(out, "# TYPE telemt_telemetry_user_enabled gauge");
+    let _ = writeln!(
+        out,
+        "telemt_telemetry_user_enabled {}",
+        if user_enabled { 1 } else { 0 }
+    );
+
+    let _ = writeln!(out, "# HELP telemt_telemetry_me_level Runtime ME telemetry level flag");
+    let _ = writeln!(out, "# TYPE telemt_telemetry_me_level gauge");
+    let _ = writeln!(
+        out,
+        "telemt_telemetry_me_level{{level=\"silent\"}} {}",
+        if matches!(telemetry.me_level, crate::config::MeTelemetryLevel::Silent) {
+            1
+        } else {
+            0
+        }
+    );
+    let _ = writeln!(
+        out,
+        "telemt_telemetry_me_level{{level=\"normal\"}} {}",
+        if matches!(telemetry.me_level, crate::config::MeTelemetryLevel::Normal) {
+            1
+        } else {
+            0
+        }
+    );
+    let _ = writeln!(
+        out,
+        "telemt_telemetry_me_level{{level=\"debug\"}} {}",
+        if matches!(telemetry.me_level, crate::config::MeTelemetryLevel::Debug) {
+            1
+        } else {
+            0
+        }
+    );
+
     let _ = writeln!(out, "# HELP telemt_connections_total Total accepted connections");
     let _ = writeln!(out, "# TYPE telemt_connections_total counter");
-    let _ = writeln!(out, "telemt_connections_total {}", stats.get_connects_all());
+    let _ = writeln!(
+        out,
+        "telemt_connections_total {}",
+        if core_enabled { stats.get_connects_all() } else { 0 }
+    );
 
     let _ = writeln!(out, "# HELP telemt_connections_bad_total Bad/rejected connections");
     let _ = writeln!(out, "# TYPE telemt_connections_bad_total counter");
-    let _ = writeln!(out, "telemt_connections_bad_total {}", stats.get_connects_bad());
+    let _ = writeln!(
+        out,
+        "telemt_connections_bad_total {}",
+        if core_enabled { stats.get_connects_bad() } else { 0 }
+    );
 
     let _ = writeln!(out, "# HELP telemt_handshake_timeouts_total Handshake timeouts");
     let _ = writeln!(out, "# TYPE telemt_handshake_timeouts_total counter");
-    let _ = writeln!(out, "telemt_handshake_timeouts_total {}", stats.get_handshake_timeouts());
+    let _ = writeln!(
+        out,
+        "telemt_handshake_timeouts_total {}",
+        if core_enabled {
+            stats.get_handshake_timeouts()
+        } else {
+            0
+        }
+    );
 
     let _ = writeln!(out, "# HELP telemt_me_keepalive_sent_total ME keepalive frames sent");
     let _ = writeln!(out, "# TYPE telemt_me_keepalive_sent_total counter");
-    let _ = writeln!(out, "telemt_me_keepalive_sent_total {}", stats.get_me_keepalive_sent());
+    let _ = writeln!(
+        out,
+        "telemt_me_keepalive_sent_total {}",
+        if me_allows_debug {
+            stats.get_me_keepalive_sent()
+        } else {
+            0
+        }
+    );
 
     let _ = writeln!(out, "# HELP telemt_me_keepalive_failed_total ME keepalive send failures");
     let _ = writeln!(out, "# TYPE telemt_me_keepalive_failed_total counter");
-    let _ = writeln!(out, "telemt_me_keepalive_failed_total {}", stats.get_me_keepalive_failed());
+    let _ = writeln!(
+        out,
+        "telemt_me_keepalive_failed_total {}",
+        if me_allows_normal {
+            stats.get_me_keepalive_failed()
+        } else {
+            0
+        }
+    );
 
     let _ = writeln!(out, "# HELP telemt_me_keepalive_pong_total ME keepalive pong replies");
     let _ = writeln!(out, "# TYPE telemt_me_keepalive_pong_total counter");
-    let _ = writeln!(out, "telemt_me_keepalive_pong_total {}", stats.get_me_keepalive_pong());
+    let _ = writeln!(
+        out,
+        "telemt_me_keepalive_pong_total {}",
+        if me_allows_debug {
+            stats.get_me_keepalive_pong()
+        } else {
+            0
+        }
+    );
 
     let _ = writeln!(out, "# HELP telemt_me_keepalive_timeout_total ME keepalive ping timeouts");
     let _ = writeln!(out, "# TYPE telemt_me_keepalive_timeout_total counter");
-    let _ = writeln!(out, "telemt_me_keepalive_timeout_total {}", stats.get_me_keepalive_timeout());
+    let _ = writeln!(
+        out,
+        "telemt_me_keepalive_timeout_total {}",
+        if me_allows_normal {
+            stats.get_me_keepalive_timeout()
+        } else {
+            0
+        }
+    );
 
     let _ = writeln!(out, "# HELP telemt_me_reconnect_attempts_total ME reconnect attempts");
     let _ = writeln!(out, "# TYPE telemt_me_reconnect_attempts_total counter");
-    let _ = writeln!(out, "telemt_me_reconnect_attempts_total {}", stats.get_me_reconnect_attempts());
+    let _ = writeln!(
+        out,
+        "telemt_me_reconnect_attempts_total {}",
+        if me_allows_normal {
+            stats.get_me_reconnect_attempts()
+        } else {
+            0
+        }
+    );
 
     let _ = writeln!(out, "# HELP telemt_me_reconnect_success_total ME reconnect successes");
     let _ = writeln!(out, "# TYPE telemt_me_reconnect_success_total counter");
-    let _ = writeln!(out, "telemt_me_reconnect_success_total {}", stats.get_me_reconnect_success());
+    let _ = writeln!(
+        out,
+        "telemt_me_reconnect_success_total {}",
+        if me_allows_normal {
+            stats.get_me_reconnect_success()
+        } else {
+            0
+        }
+    );
 
     let _ = writeln!(out, "# HELP telemt_me_crc_mismatch_total ME CRC mismatches");
     let _ = writeln!(out, "# TYPE telemt_me_crc_mismatch_total counter");
-    let _ = writeln!(out, "telemt_me_crc_mismatch_total {}", stats.get_me_crc_mismatch());
+    let _ = writeln!(
+        out,
+        "telemt_me_crc_mismatch_total {}",
+        if me_allows_normal {
+            stats.get_me_crc_mismatch()
+        } else {
+            0
+        }
+    );
 
     let _ = writeln!(out, "# HELP telemt_me_seq_mismatch_total ME sequence mismatches");
     let _ = writeln!(out, "# TYPE telemt_me_seq_mismatch_total counter");
-    let _ = writeln!(out, "telemt_me_seq_mismatch_total {}", stats.get_me_seq_mismatch());
+    let _ = writeln!(
+        out,
+        "telemt_me_seq_mismatch_total {}",
+        if me_allows_normal {
+            stats.get_me_seq_mismatch()
+        } else {
+            0
+        }
+    );
 
     let _ = writeln!(out, "# HELP telemt_me_route_drop_no_conn_total ME route drops: no conn");
     let _ = writeln!(out, "# TYPE telemt_me_route_drop_no_conn_total counter");
-    let _ = writeln!(out, "telemt_me_route_drop_no_conn_total {}", stats.get_me_route_drop_no_conn());
+    let _ = writeln!(
+        out,
+        "telemt_me_route_drop_no_conn_total {}",
+        if me_allows_normal {
+            stats.get_me_route_drop_no_conn()
+        } else {
+            0
+        }
+    );
 
     let _ = writeln!(out, "# HELP telemt_me_route_drop_channel_closed_total ME route drops: channel closed");
     let _ = writeln!(out, "# TYPE telemt_me_route_drop_channel_closed_total counter");
-    let _ = writeln!(out, "telemt_me_route_drop_channel_closed_total {}", stats.get_me_route_drop_channel_closed());
+    let _ = writeln!(
+        out,
+        "telemt_me_route_drop_channel_closed_total {}",
+        if me_allows_normal {
+            stats.get_me_route_drop_channel_closed()
+        } else {
+            0
+        }
+    );
 
     let _ = writeln!(out, "# HELP telemt_me_route_drop_queue_full_total ME route drops: queue full");
     let _ = writeln!(out, "# TYPE telemt_me_route_drop_queue_full_total counter");
-    let _ = writeln!(out, "telemt_me_route_drop_queue_full_total {}", stats.get_me_route_drop_queue_full());
+    let _ = writeln!(
+        out,
+        "telemt_me_route_drop_queue_full_total {}",
+        if me_allows_normal {
+            stats.get_me_route_drop_queue_full()
+        } else {
+            0
+        }
+    );
+
+    let _ = writeln!(
+        out,
+        "# HELP telemt_me_route_drop_queue_full_profile_total ME route drops: queue full by adaptive profile"
+    );
+    let _ = writeln!(
+        out,
+        "# TYPE telemt_me_route_drop_queue_full_profile_total counter"
+    );
+    let _ = writeln!(
+        out,
+        "telemt_me_route_drop_queue_full_profile_total{{profile=\"base\"}} {}",
+        if me_allows_normal {
+            stats.get_me_route_drop_queue_full_base()
+        } else {
+            0
+        }
+    );
+    let _ = writeln!(
+        out,
+        "telemt_me_route_drop_queue_full_profile_total{{profile=\"high\"}} {}",
+        if me_allows_normal {
+            stats.get_me_route_drop_queue_full_high()
+        } else {
+            0
+        }
+    );
+
+    let _ = writeln!(
+        out,
+        "# HELP telemt_me_socks_kdf_policy_total SOCKS KDF policy outcomes"
+    );
+    let _ = writeln!(out, "# TYPE telemt_me_socks_kdf_policy_total counter");
+    let _ = writeln!(
+        out,
+        "telemt_me_socks_kdf_policy_total{{policy=\"strict\",outcome=\"reject\"}} {}",
+        if me_allows_normal {
+            stats.get_me_socks_kdf_strict_reject()
+        } else {
+            0
+        }
+    );
+    let _ = writeln!(
+        out,
+        "telemt_me_socks_kdf_policy_total{{policy=\"compat\",outcome=\"fallback\"}} {}",
+        if me_allows_debug {
+            stats.get_me_socks_kdf_compat_fallback()
+        } else {
+            0
+        }
+    );
 
     let _ = writeln!(out, "# HELP telemt_secure_padding_invalid_total Invalid secure frame lengths");
     let _ = writeln!(out, "# TYPE telemt_secure_padding_invalid_total counter");
-    let _ = writeln!(out, "telemt_secure_padding_invalid_total {}", stats.get_secure_padding_invalid());
+    let _ = writeln!(
+        out,
+        "telemt_secure_padding_invalid_total {}",
+        if me_allows_normal {
+            stats.get_secure_padding_invalid()
+        } else {
+            0
+        }
+    );
 
     let _ = writeln!(out, "# HELP telemt_desync_total Total crypto-desync detections");
     let _ = writeln!(out, "# TYPE telemt_desync_total counter");
-    let _ = writeln!(out, "telemt_desync_total {}", stats.get_desync_total());
+    let _ = writeln!(
+        out,
+        "telemt_desync_total {}",
+        if me_allows_normal {
+            stats.get_desync_total()
+        } else {
+            0
+        }
+    );
 
     let _ = writeln!(out, "# HELP telemt_desync_full_logged_total Full forensic desync logs emitted");
     let _ = writeln!(out, "# TYPE telemt_desync_full_logged_total counter");
-    let _ = writeln!(out, "telemt_desync_full_logged_total {}", stats.get_desync_full_logged());
+    let _ = writeln!(
+        out,
+        "telemt_desync_full_logged_total {}",
+        if me_allows_normal {
+            stats.get_desync_full_logged()
+        } else {
+            0
+        }
+    );
 
     let _ = writeln!(out, "# HELP telemt_desync_suppressed_total Suppressed desync forensic events");
     let _ = writeln!(out, "# TYPE telemt_desync_suppressed_total counter");
-    let _ = writeln!(out, "telemt_desync_suppressed_total {}", stats.get_desync_suppressed());
+    let _ = writeln!(
+        out,
+        "telemt_desync_suppressed_total {}",
+        if me_allows_normal {
+            stats.get_desync_suppressed()
+        } else {
+            0
+        }
+    );
 
     let _ = writeln!(out, "# HELP telemt_desync_frames_bucket_total Desync count by frames_ok bucket");
     let _ = writeln!(out, "# TYPE telemt_desync_frames_bucket_total counter");
     let _ = writeln!(
         out,
         "telemt_desync_frames_bucket_total{{bucket=\"0\"}} {}",
-        stats.get_desync_frames_bucket_0()
+        if me_allows_normal {
+            stats.get_desync_frames_bucket_0()
+        } else {
+            0
+        }
     );
     let _ = writeln!(
         out,
         "telemt_desync_frames_bucket_total{{bucket=\"1_2\"}} {}",
-        stats.get_desync_frames_bucket_1_2()
+        if me_allows_normal {
+            stats.get_desync_frames_bucket_1_2()
+        } else {
+            0
+        }
     );
     let _ = writeln!(
         out,
         "telemt_desync_frames_bucket_total{{bucket=\"3_10\"}} {}",
-        stats.get_desync_frames_bucket_3_10()
+        if me_allows_normal {
+            stats.get_desync_frames_bucket_3_10()
+        } else {
+            0
+        }
     );
     let _ = writeln!(
         out,
         "telemt_desync_frames_bucket_total{{bucket=\"gt_10\"}} {}",
-        stats.get_desync_frames_bucket_gt_10()
+        if me_allows_normal {
+            stats.get_desync_frames_bucket_gt_10()
+        } else {
+            0
+        }
     );
 
     let _ = writeln!(out, "# HELP telemt_pool_swap_total Successful ME pool swaps");
     let _ = writeln!(out, "# TYPE telemt_pool_swap_total counter");
-    let _ = writeln!(out, "telemt_pool_swap_total {}", stats.get_pool_swap_total());
+    let _ = writeln!(
+        out,
+        "telemt_pool_swap_total {}",
+        if me_allows_debug {
+            stats.get_pool_swap_total()
+        } else {
+            0
+        }
+    );
 
     let _ = writeln!(out, "# HELP telemt_pool_drain_active Active draining ME writers");
     let _ = writeln!(out, "# TYPE telemt_pool_drain_active gauge");
-    let _ = writeln!(out, "telemt_pool_drain_active {}", stats.get_pool_drain_active());
+    let _ = writeln!(
+        out,
+        "telemt_pool_drain_active {}",
+        if me_allows_debug {
+            stats.get_pool_drain_active()
+        } else {
+            0
+        }
+    );
 
     let _ = writeln!(out, "# HELP telemt_pool_force_close_total Forced close events for draining writers");
     let _ = writeln!(out, "# TYPE telemt_pool_force_close_total counter");
     let _ = writeln!(
         out,
         "telemt_pool_force_close_total {}",
-        stats.get_pool_force_close_total()
+        if me_allows_normal {
+            stats.get_pool_force_close_total()
+        } else {
+            0
+        }
     );
 
     let _ = writeln!(out, "# HELP telemt_pool_stale_pick_total Stale writer fallback picks for new binds");
@@ -196,7 +513,132 @@ fn render_metrics(stats: &Stats) -> String {
     let _ = writeln!(
         out,
         "telemt_pool_stale_pick_total {}",
-        stats.get_pool_stale_pick_total()
+        if me_allows_normal {
+            stats.get_pool_stale_pick_total()
+        } else {
+            0
+        }
+    );
+
+    let _ = writeln!(out, "# HELP telemt_me_writer_removed_total Total ME writer removals");
+    let _ = writeln!(out, "# TYPE telemt_me_writer_removed_total counter");
+    let _ = writeln!(
+        out,
+        "telemt_me_writer_removed_total {}",
+        if me_allows_debug {
+            stats.get_me_writer_removed_total()
+        } else {
+            0
+        }
+    );
+
+    let _ = writeln!(
+        out,
+        "# HELP telemt_me_writer_removed_unexpected_total Unexpected ME writer removals that triggered refill"
+    );
+    let _ = writeln!(out, "# TYPE telemt_me_writer_removed_unexpected_total counter");
+    let _ = writeln!(
+        out,
+        "telemt_me_writer_removed_unexpected_total {}",
+        if me_allows_normal {
+            stats.get_me_writer_removed_unexpected_total()
+        } else {
+            0
+        }
+    );
+
+    let _ = writeln!(out, "# HELP telemt_me_refill_triggered_total Immediate ME refill runs started");
+    let _ = writeln!(out, "# TYPE telemt_me_refill_triggered_total counter");
+    let _ = writeln!(
+        out,
+        "telemt_me_refill_triggered_total {}",
+        if me_allows_debug {
+            stats.get_me_refill_triggered_total()
+        } else {
+            0
+        }
+    );
+
+    let _ = writeln!(
+        out,
+        "# HELP telemt_me_refill_skipped_inflight_total Immediate ME refill skips due to inflight dedup"
+    );
+    let _ = writeln!(out, "# TYPE telemt_me_refill_skipped_inflight_total counter");
+    let _ = writeln!(
+        out,
+        "telemt_me_refill_skipped_inflight_total {}",
+        if me_allows_debug {
+            stats.get_me_refill_skipped_inflight_total()
+        } else {
+            0
+        }
+    );
+
+    let _ = writeln!(out, "# HELP telemt_me_refill_failed_total Immediate ME refill failures");
+    let _ = writeln!(out, "# TYPE telemt_me_refill_failed_total counter");
+    let _ = writeln!(
+        out,
+        "telemt_me_refill_failed_total {}",
+        if me_allows_normal {
+            stats.get_me_refill_failed_total()
+        } else {
+            0
+        }
+    );
+
+    let _ = writeln!(
+        out,
+        "# HELP telemt_me_writer_restored_same_endpoint_total Refilled ME writer restored on the same endpoint"
+    );
+    let _ = writeln!(out, "# TYPE telemt_me_writer_restored_same_endpoint_total counter");
+    let _ = writeln!(
+        out,
+        "telemt_me_writer_restored_same_endpoint_total {}",
+        if me_allows_normal {
+            stats.get_me_writer_restored_same_endpoint_total()
+        } else {
+            0
+        }
+    );
+
+    let _ = writeln!(
+        out,
+        "# HELP telemt_me_writer_restored_fallback_total Refilled ME writer restored via fallback endpoint"
+    );
+    let _ = writeln!(out, "# TYPE telemt_me_writer_restored_fallback_total counter");
+    let _ = writeln!(
+        out,
+        "telemt_me_writer_restored_fallback_total {}",
+        if me_allows_normal {
+            stats.get_me_writer_restored_fallback_total()
+        } else {
+            0
+        }
+    );
+
+    let unresolved_writer_losses = if me_allows_normal {
+        stats
+            .get_me_writer_removed_unexpected_total()
+            .saturating_sub(
+                stats
+                    .get_me_writer_restored_same_endpoint_total()
+                    .saturating_add(stats.get_me_writer_restored_fallback_total()),
+            )
+    } else {
+        0
+    };
+    let _ = writeln!(
+        out,
+        "# HELP telemt_me_writer_removed_unexpected_minus_restored_total Unexpected writer removals not yet compensated by restore"
+    );
+    let _ = writeln!(
+        out,
+        "# TYPE telemt_me_writer_removed_unexpected_minus_restored_total gauge"
+    );
+    let _ = writeln!(
+        out,
+        "telemt_me_writer_removed_unexpected_minus_restored_total {}",
+        unresolved_writer_losses
     );
 
     let _ = writeln!(out, "# HELP telemt_user_connections_total Per-user total connections");
@@ -211,16 +653,63 @@ fn render_metrics(stats: &Stats) -> String {
     let _ = writeln!(out, "# TYPE telemt_user_msgs_from_client counter");
     let _ = writeln!(out, "# HELP telemt_user_msgs_to_client Per-user messages sent");
     let _ = writeln!(out, "# TYPE telemt_user_msgs_to_client counter");
+    let _ = writeln!(
+        out,
+        "# HELP telemt_telemetry_user_series_suppressed User-labeled metric series suppression flag"
+    );
+    let _ = writeln!(out, "# TYPE telemt_telemetry_user_series_suppressed gauge");
+    let _ = writeln!(
+        out,
+        "telemt_telemetry_user_series_suppressed {}",
+        if user_enabled { 0 } else { 1 }
+    );
 
-    for entry in stats.iter_user_stats() {
-        let user = entry.key();
-        let s = entry.value();
-        let _ = writeln!(out, "telemt_user_connections_total{{user=\"{}\"}} {}", user, s.connects.load(std::sync::atomic::Ordering::Relaxed));
-        let _ = writeln!(out, "telemt_user_connections_current{{user=\"{}\"}} {}", user, s.curr_connects.load(std::sync::atomic::Ordering::Relaxed));
-        let _ = writeln!(out, "telemt_user_octets_from_client{{user=\"{}\"}} {}", user, s.octets_from_client.load(std::sync::atomic::Ordering::Relaxed));
-        let _ = writeln!(out, "telemt_user_octets_to_client{{user=\"{}\"}} {}", user, s.octets_to_client.load(std::sync::atomic::Ordering::Relaxed));
-        let _ = writeln!(out, "telemt_user_msgs_from_client{{user=\"{}\"}} {}", user, s.msgs_from_client.load(std::sync::atomic::Ordering::Relaxed));
-        let _ = writeln!(out, "telemt_user_msgs_to_client{{user=\"{}\"}} {}", user, s.msgs_to_client.load(std::sync::atomic::Ordering::Relaxed));
+    if user_enabled {
+        for entry in stats.iter_user_stats() {
+            let user = entry.key();
+            let s = entry.value();
+            let _ = writeln!(out, "telemt_user_connections_total{{user=\"{}\"}} {}", user, s.connects.load(std::sync::atomic::Ordering::Relaxed));
+            let _ = writeln!(out, "telemt_user_connections_current{{user=\"{}\"}} {}", user, s.curr_connects.load(std::sync::atomic::Ordering::Relaxed));
+            let _ = writeln!(out, "telemt_user_octets_from_client{{user=\"{}\"}} {}", user, s.octets_from_client.load(std::sync::atomic::Ordering::Relaxed));
+            let _ = writeln!(out, "telemt_user_octets_to_client{{user=\"{}\"}} {}", user, s.octets_to_client.load(std::sync::atomic::Ordering::Relaxed));
+            let _ = writeln!(out, "telemt_user_msgs_from_client{{user=\"{}\"}} {}", user, s.msgs_from_client.load(std::sync::atomic::Ordering::Relaxed));
+            let _ = writeln!(out, "telemt_user_msgs_to_client{{user=\"{}\"}} {}", user, s.msgs_to_client.load(std::sync::atomic::Ordering::Relaxed));
+        }
+
+        let ip_stats = ip_tracker.get_stats().await;
+        let ip_counts: HashMap<String, usize> = ip_stats
+            .into_iter()
+            .map(|(user, count, _)| (user, count))
+            .collect();
+
+        let mut unique_users = BTreeSet::new();
+        unique_users.extend(config.access.user_max_unique_ips.keys().cloned());
+        unique_users.extend(ip_counts.keys().cloned());
+
+        let _ = writeln!(out, "# HELP telemt_user_unique_ips_current Per-user current number of unique active IPs");
+        let _ = writeln!(out, "# TYPE telemt_user_unique_ips_current gauge");
+        let _ = writeln!(out, "# HELP telemt_user_unique_ips_limit Per-user configured unique IP limit (0 means unlimited)");
+        let _ = writeln!(out, "# TYPE telemt_user_unique_ips_limit gauge");
+        let _ = writeln!(out, "# HELP telemt_user_unique_ips_utilization Per-user unique IP usage ratio (0 for unlimited)");
+        let _ = writeln!(out, "# TYPE telemt_user_unique_ips_utilization gauge");
+
+        for user in unique_users {
+            let current = ip_counts.get(&user).copied().unwrap_or(0);
+            let limit = config.access.user_max_unique_ips.get(&user).copied().unwrap_or(0);
+            let utilization = if limit > 0 {
+                current as f64 / limit as f64
+            } else {
+                0.0
+            };
+            let _ = writeln!(out, "telemt_user_unique_ips_current{{user=\"{}\"}} {}", user, current);
+            let _ = writeln!(out, "telemt_user_unique_ips_limit{{user=\"{}\"}} {}", user, limit);
+            let _ = writeln!(
+                out,
+                "telemt_user_unique_ips_utilization{{user=\"{}\"}} {:.6}",
+                user,
+                utilization
+            );
+        }
     }
 
     out
@@ -229,11 +718,19 @@ fn render_metrics(stats: &Stats) -> String {
 #[cfg(test)]
 mod tests {
     use super::*;
+    use std::net::IpAddr;
     use http_body_util::BodyExt;
 
-    #[test]
-    fn test_render_metrics_format() {
+    #[tokio::test]
+    async fn test_render_metrics_format() {
         let stats = Arc::new(Stats::new());
+        let tracker = UserIpTracker::new();
+        let mut config = ProxyConfig::default();
+        config
+            .access
+            .user_max_unique_ips
+            .insert("alice".to_string(), 4);
+
         stats.increment_connects_all();
         stats.increment_connects_all();
         stats.increment_connects_bad();
@@ -245,8 +742,12 @@ mod tests {
         stats.increment_user_msgs_from("alice");
         stats.increment_user_msgs_to("alice");
         stats.increment_user_msgs_to("alice");
+        tracker
+            .check_and_add("alice", "203.0.113.10".parse().unwrap())
+            .await
+            .unwrap();
 
-        let output = render_metrics(&stats);
+        let output = render_metrics(&stats, &config, &tracker).await;
 
         assert!(output.contains("telemt_connections_total 2"));
         assert!(output.contains("telemt_connections_bad_total 1"));
@@ -257,31 +758,48 @@ mod tests {
         assert!(output.contains("telemt_user_octets_to_client{user=\"alice\"} 2048"));
         assert!(output.contains("telemt_user_msgs_from_client{user=\"alice\"} 1"));
         assert!(output.contains("telemt_user_msgs_to_client{user=\"alice\"} 2"));
+        assert!(output.contains("telemt_user_unique_ips_current{user=\"alice\"} 1"));
+        assert!(output.contains("telemt_user_unique_ips_limit{user=\"alice\"} 4"));
+        assert!(output.contains("telemt_user_unique_ips_utilization{user=\"alice\"} 0.250000"));
     }
 
-    #[test]
-    fn test_render_empty_stats() {
+    #[tokio::test]
+    async fn test_render_empty_stats() {
         let stats = Stats::new();
-        let output = render_metrics(&stats);
+        let tracker = UserIpTracker::new();
+        let config = ProxyConfig::default();
+        let output = render_metrics(&stats, &config, &tracker).await;
         assert!(output.contains("telemt_connections_total 0"));
         assert!(output.contains("telemt_connections_bad_total 0"));
         assert!(output.contains("telemt_handshake_timeouts_total 0"));
         assert!(!output.contains("user="));
     }
 
-    #[test]
-    fn test_render_has_type_annotations() {
+    #[tokio::test]
+    async fn test_render_has_type_annotations() {
         let stats = Stats::new();
-        let output = render_metrics(&stats);
+        let tracker = UserIpTracker::new();
+        let config = ProxyConfig::default();
+        let output = render_metrics(&stats, &config, &tracker).await;
         assert!(output.contains("# TYPE telemt_uptime_seconds gauge"));
         assert!(output.contains("# TYPE telemt_connections_total counter"));
         assert!(output.contains("# TYPE telemt_connections_bad_total counter"));
         assert!(output.contains("# TYPE telemt_handshake_timeouts_total counter"));
+        assert!(output.contains("# TYPE telemt_me_writer_removed_total counter"));
+        assert!(output.contains(
+            "# TYPE telemt_me_writer_removed_unexpected_minus_restored_total gauge"
+        ));
+        assert!(output.contains("# TYPE telemt_user_unique_ips_current gauge"));
+        assert!(output.contains("# TYPE telemt_user_unique_ips_limit gauge"));
+        assert!(output.contains("# TYPE telemt_user_unique_ips_utilization gauge"));
     }
 
     #[tokio::test]
     async fn test_endpoint_integration() {
         let stats = Arc::new(Stats::new());
+        let beobachten = Arc::new(BeobachtenStore::new());
+        let tracker = UserIpTracker::new();
+        let mut config = ProxyConfig::default();
         stats.increment_connects_all();
         stats.increment_connects_all();
         stats.increment_connects_all();
@@ -290,16 +808,38 @@ mod tests {
             .uri("/metrics")
             .body(())
             .unwrap();
-        let resp = handle(req, &stats).unwrap();
+        let resp = handle(req, &stats, &beobachten, &tracker, &config).await.unwrap();
         assert_eq!(resp.status(), StatusCode::OK);
         let body = resp.into_body().collect().await.unwrap().to_bytes();
         assert!(std::str::from_utf8(body.as_ref()).unwrap().contains("telemt_connections_total 3"));
 
+        config.general.beobachten = true;
+        config.general.beobachten_minutes = 10;
+        beobachten.record(
+            "TLS-scanner",
+            "203.0.113.10".parse::<IpAddr>().unwrap(),
+            Duration::from_secs(600),
+        );
+        let req_beob = Request::builder()
+            .uri("/beobachten")
+            .body(())
+            .unwrap();
+        let resp_beob = handle(req_beob, &stats, &beobachten, &tracker, &config)
+            .await
+            .unwrap();
+        assert_eq!(resp_beob.status(), StatusCode::OK);
+        let body_beob = resp_beob.into_body().collect().await.unwrap().to_bytes();
+        let beob_text = std::str::from_utf8(body_beob.as_ref()).unwrap();
+        assert!(beob_text.contains("[TLS-scanner]"));
+        assert!(beob_text.contains("203.0.113.10-1"));
+
         let req404 = Request::builder()
             .uri("/other")
             .body(())
             .unwrap();
-        let resp404 = handle(req404, &stats).unwrap();
+        let resp404 = handle(req404, &stats, &beobachten, &tracker, &config)
+            .await
+            .unwrap();
         assert_eq!(resp404.status(), StatusCode::NOT_FOUND);
     }
 }

src/network/dns_overrides.rs

@@ -0,0 +1,197 @@
+//! Runtime DNS overrides for `host:port` targets.
+
+use std::collections::HashMap;
+use std::net::{IpAddr, Ipv6Addr, SocketAddr};
+use std::sync::{OnceLock, RwLock};
+
+use crate::error::{ProxyError, Result};
+
+type OverrideMap = HashMap<(String, u16), IpAddr>;
+
+static DNS_OVERRIDES: OnceLock<RwLock<OverrideMap>> = OnceLock::new();
+
+fn overrides_store() -> &'static RwLock<OverrideMap> {
+    DNS_OVERRIDES.get_or_init(|| RwLock::new(HashMap::new()))
+}
+
+fn parse_ip_spec(ip_spec: &str) -> Result<IpAddr> {
+    if ip_spec.starts_with('[') && ip_spec.ends_with(']') {
+        let inner = &ip_spec[1..ip_spec.len() - 1];
+        let ipv6 = inner.parse::<Ipv6Addr>().map_err(|_| {
+            ProxyError::Config(format!(
+                "network.dns_overrides IPv6 override is invalid: '{ip_spec}'"
+            ))
+        })?;
+        return Ok(IpAddr::V6(ipv6));
+    }
+
+    let ip = ip_spec.parse::<IpAddr>().map_err(|_| {
+        ProxyError::Config(format!(
+            "network.dns_overrides IP is invalid: '{ip_spec}'"
+        ))
+    })?;
+    if matches!(ip, IpAddr::V6(_)) {
+        return Err(ProxyError::Config(format!(
+            "network.dns_overrides IPv6 must be bracketed: '{ip_spec}'"
+        )));
+    }
+    Ok(ip)
+}
+
+fn parse_entry(entry: &str) -> Result<((String, u16), IpAddr)> {
+    let trimmed = entry.trim();
+    if trimmed.is_empty() {
+        return Err(ProxyError::Config(
+            "network.dns_overrides entry cannot be empty".to_string(),
+        ));
+    }
+
+    let first_sep = trimmed.find(':').ok_or_else(|| {
+        ProxyError::Config(format!(
+            "network.dns_overrides entry must use host:port:ip format: '{trimmed}'"
+        ))
+    })?;
+    let second_sep = trimmed[first_sep + 1..]
+        .find(':')
+        .map(|idx| first_sep + 1 + idx)
+        .ok_or_else(|| {
+            ProxyError::Config(format!(
+                "network.dns_overrides entry must use host:port:ip format: '{trimmed}'"
+            ))
+        })?;
+
+    let host = trimmed[..first_sep].trim();
+    let port_str = trimmed[first_sep + 1..second_sep].trim();
+    let ip_str = trimmed[second_sep + 1..].trim();
+
+    if host.is_empty() {
+        return Err(ProxyError::Config(format!(
+            "network.dns_overrides host cannot be empty: '{trimmed}'"
+        )));
+    }
+    if host.contains(':') {
+        return Err(ProxyError::Config(format!(
+            "network.dns_overrides host must be a domain name without ':' in this format: '{trimmed}'"
+        )));
+    }
+
+    let port = port_str.parse::<u16>().map_err(|_| {
+        ProxyError::Config(format!(
+            "network.dns_overrides port is invalid: '{trimmed}'"
+        ))
+    })?;
+    let ip = parse_ip_spec(ip_str)?;
+
+    Ok(((host.to_ascii_lowercase(), port), ip))
+}
+
+fn parse_entries(entries: &[String]) -> Result<OverrideMap> {
+    let mut parsed = HashMap::new();
+    for entry in entries {
+        let (key, ip) = parse_entry(entry)?;
+        parsed.insert(key, ip);
+    }
+    Ok(parsed)
+}
+
+/// Validate `network.dns_overrides` entries without updating runtime state.
+pub fn validate_entries(entries: &[String]) -> Result<()> {
+    let _ = parse_entries(entries)?;
+    Ok(())
+}
+
+/// Replace runtime DNS overrides with a new validated snapshot.
+pub fn install_entries(entries: &[String]) -> Result<()> {
+    let parsed = parse_entries(entries)?;
+    let mut guard = overrides_store()
+        .write()
+        .map_err(|_| ProxyError::Config("network.dns_overrides runtime lock is poisoned".to_string()))?;
+    *guard = parsed;
+    Ok(())
+}
+
+/// Resolve a hostname override for `(host, port)` if present.
+pub fn resolve(host: &str, port: u16) -> Option<IpAddr> {
+    let key = (host.to_ascii_lowercase(), port);
+    overrides_store()
+        .read()
+        .ok()
+        .and_then(|guard| guard.get(&key).copied())
+}
+
+/// Resolve a hostname override and construct a socket address when present.
+pub fn resolve_socket_addr(host: &str, port: u16) -> Option<SocketAddr> {
+    resolve(host, port).map(|ip| SocketAddr::new(ip, port))
+}
+
+/// Parse a runtime endpoint in `host:port` format.
+///
+/// Supports:
+/// - `example.com:443`
+/// - `[2001:db8::1]:443`
+pub fn split_host_port(endpoint: &str) -> Option<(String, u16)> {
+    if endpoint.starts_with('[') {
+        let bracket_end = endpoint.find(']')?;
+        if endpoint.as_bytes().get(bracket_end + 1) != Some(&b':') {
+            return None;
+        }
+        let host = endpoint[1..bracket_end].trim();
+        let port = endpoint[bracket_end + 2..].trim().parse::<u16>().ok()?;
+        if host.is_empty() {
+            return None;
+        }
+        return Some((host.to_ascii_lowercase(), port));
+    }
+
+    let split_idx = endpoint.rfind(':')?;
+    let host = endpoint[..split_idx].trim();
+    let port = endpoint[split_idx + 1..].trim().parse::<u16>().ok()?;
+    if host.is_empty() || host.contains(':') {
+        return None;
+    }
+
+    Some((host.to_ascii_lowercase(), port))
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn validate_accepts_ipv4_and_bracketed_ipv6() {
+        let entries = vec![
+            "example.com:443:127.0.0.1".to_string(),
+            "example.net:8443:[2001:db8::10]".to_string(),
+        ];
+        assert!(validate_entries(&entries).is_ok());
+    }
+
+    #[test]
+    fn validate_rejects_unbracketed_ipv6() {
+        let entries = vec!["example.net:443:2001:db8::10".to_string()];
+        let err = validate_entries(&entries).unwrap_err().to_string();
+        assert!(err.contains("must be bracketed"));
+    }
+
+    #[test]
+    fn install_and_resolve_are_case_insensitive_for_host() {
+        let entries = vec!["MyPetrovich.ru:8443:127.0.0.1".to_string()];
+        install_entries(&entries).unwrap();
+
+        let resolved = resolve("mypetrovich.ru", 8443);
+        assert_eq!(resolved, Some("127.0.0.1".parse().unwrap()));
+    }
+
+    #[test]
+    fn split_host_port_parses_supported_shapes() {
+        assert_eq!(
+            split_host_port("example.com:443"),
+            Some(("example.com".to_string(), 443))
+        );
+        assert_eq!(
+            split_host_port("[2001:db8::1]:443"),
+            Some(("2001:db8::1".to_string(), 443))
+        );
+        assert_eq!(split_host_port("2001:db8::1:443"), None);
+    }
+}

src/network/mod.rs

@@ -1,3 +1,4 @@
+pub mod dns_overrides;
 pub mod probe;
 pub mod stun;
 

src/network/probe.rs

@@ -1,12 +1,16 @@
 #![allow(dead_code)]
 
+use std::collections::HashMap;
 use std::net::{IpAddr, Ipv4Addr, Ipv6Addr, SocketAddr, UdpSocket};
+use std::time::Duration;
 
-use tracing::{info, warn};
+use tokio::task::JoinSet;
+use tokio::time::timeout;
+use tracing::{debug, info, warn};
 
 use crate::config::NetworkConfig;
 use crate::error::Result;
-use crate::network::stun::{stun_probe_dual, DualStunResult, IpFamily};
+use crate::network::stun::{stun_probe_dual, DualStunResult, IpFamily, StunProbeResult};
 
 #[derive(Debug, Clone, Default)]
 pub struct NetworkProbe {
@@ -49,7 +53,13 @@ impl NetworkDecision {
     }
 }
 
-pub async fn run_probe(config: &NetworkConfig, stun_addr: Option<String>, nat_probe: bool) -> Result<NetworkProbe> {
+const STUN_BATCH_TIMEOUT: Duration = Duration::from_secs(5);
+
+pub async fn run_probe(
+    config: &NetworkConfig,
+    nat_probe: bool,
+    stun_nat_probe_concurrency: usize,
+) -> Result<NetworkProbe> {
     let mut probe = NetworkProbe::default();
 
     probe.detected_ipv4 = detect_local_ip_v4();
@@ -58,21 +68,38 @@ pub async fn run_probe(config: &NetworkConfig, stun_addr: Option<String>, nat_pr
     probe.ipv4_is_bogon = probe.detected_ipv4.map(is_bogon_v4).unwrap_or(false);
     probe.ipv6_is_bogon = probe.detected_ipv6.map(is_bogon_v6).unwrap_or(false);
 
-    let stun_server = stun_addr.unwrap_or_else(|| "stun.l.google.com:19302".to_string());
-    let stun_res = if nat_probe {
-        match stun_probe_dual(&stun_server).await {
-            Ok(res) => res,
-            Err(e) => {
-                warn!(error = %e, "STUN probe failed, continuing without reflection");
-                DualStunResult::default()
-            }
+    let stun_res = if nat_probe && config.stun_use {
+        let servers = collect_stun_servers(config);
+        if servers.is_empty() {
+            warn!("STUN probe is enabled but network.stun_servers is empty");
+            DualStunResult::default()
+        } else {
+            probe_stun_servers_parallel(
+                &servers,
+                stun_nat_probe_concurrency.max(1),
+            )
+            .await
         }
+    } else if nat_probe {
+        info!("STUN probe is disabled by network.stun_use=false");
+        DualStunResult::default()
     } else {
         DualStunResult::default()
     };
     probe.reflected_ipv4 = stun_res.v4.map(|r| r.reflected_addr);
     probe.reflected_ipv6 = stun_res.v6.map(|r| r.reflected_addr);
 
+    // If STUN is blocked but IPv4 is private, try HTTP public-IP fallback.
+    if nat_probe
+        && probe.reflected_ipv4.is_none()
+        && probe.detected_ipv4.map(is_bogon_v4).unwrap_or(false)
+    {
+        if let Some(public_ip) = detect_public_ipv4_http(&config.http_ip_detect_urls).await {
+            probe.reflected_ipv4 = Some(SocketAddr::new(IpAddr::V4(public_ip), 0));
+            info!(public_ip = %public_ip, "STUN unavailable, using HTTP public IPv4 fallback");
+        }
+    }
+
     probe.ipv4_nat_detected = match (probe.detected_ipv4, probe.reflected_ipv4) {
         (Some(det), Some(reflected)) => det != reflected.ip(),
         _ => false,
@@ -94,6 +121,111 @@ pub async fn run_probe(config: &NetworkConfig, stun_addr: Option<String>, nat_pr
     Ok(probe)
 }
 
+async fn detect_public_ipv4_http(urls: &[String]) -> Option<Ipv4Addr> {
+    let client = reqwest::Client::builder()
+        .timeout(Duration::from_secs(3))
+        .build()
+        .ok()?;
+
+    for url in urls {
+        let response = match client.get(url).send().await {
+            Ok(response) => response,
+            Err(_) => continue,
+        };
+
+        let body = match response.text().await {
+            Ok(body) => body,
+            Err(_) => continue,
+        };
+
+        let Ok(ip) = body.trim().parse::<Ipv4Addr>() else {
+            continue;
+        };
+        if !is_bogon_v4(ip) {
+            return Some(ip);
+        }
+    }
+
+    None
+}
+
+fn collect_stun_servers(config: &NetworkConfig) -> Vec<String> {
+    let mut out = Vec::new();
+    for s in &config.stun_servers {
+        if !s.is_empty() && !out.contains(s) {
+            out.push(s.clone());
+        }
+    }
+    out
+}
+
+async fn probe_stun_servers_parallel(
+    servers: &[String],
+    concurrency: usize,
+) -> DualStunResult {
+    let mut join_set = JoinSet::new();
+    let mut next_idx = 0usize;
+    let mut best_v4_by_ip: HashMap<IpAddr, (usize, StunProbeResult)> = HashMap::new();
+    let mut best_v6_by_ip: HashMap<IpAddr, (usize, StunProbeResult)> = HashMap::new();
+
+    while next_idx < servers.len() || !join_set.is_empty() {
+        while next_idx < servers.len() && join_set.len() < concurrency {
+            let stun_addr = servers[next_idx].clone();
+            next_idx += 1;
+            join_set.spawn(async move {
+                let res = timeout(STUN_BATCH_TIMEOUT, stun_probe_dual(&stun_addr)).await;
+                (stun_addr, res)
+            });
+        }
+
+        let Some(task) = join_set.join_next().await else {
+            break;
+        };
+
+        match task {
+            Ok((stun_addr, Ok(Ok(result)))) => {
+                if let Some(v4) = result.v4 {
+                    let entry = best_v4_by_ip.entry(v4.reflected_addr.ip()).or_insert((0, v4));
+                    entry.0 += 1;
+                }
+                if let Some(v6) = result.v6 {
+                    let entry = best_v6_by_ip.entry(v6.reflected_addr.ip()).or_insert((0, v6));
+                    entry.0 += 1;
+                }
+                if result.v4.is_some() || result.v6.is_some() {
+                    debug!(stun = %stun_addr, "STUN server responded within probe timeout");
+                }
+            }
+            Ok((stun_addr, Ok(Err(e)))) => {
+                debug!(error = %e, stun = %stun_addr, "STUN probe failed");
+            }
+            Ok((stun_addr, Err(_))) => {
+                debug!(stun = %stun_addr, "STUN probe timeout");
+            }
+            Err(e) => {
+                debug!(error = %e, "STUN probe task join failed");
+            }
+        }
+    }
+
+    let mut out = DualStunResult::default();
+    if let Some((_, best)) = best_v4_by_ip
+        .into_values()
+        .max_by_key(|(count, _)| *count)
+    {
+        info!("STUN-Quorum reached, IP: {}", best.reflected_addr.ip());
+        out.v4 = Some(best);
+    }
+    if let Some((_, best)) = best_v6_by_ip
+        .into_values()
+        .max_by_key(|(count, _)| *count)
+    {
+        info!("STUN-Quorum reached, IP: {}", best.reflected_addr.ip());
+        out.v6 = Some(best);
+    }
+    out
+}
+
 pub fn decide_network_capabilities(config: &NetworkConfig, probe: &NetworkProbe) -> NetworkDecision {
     let ipv4_dc = config.ipv4 && probe.detected_ipv4.is_some();
     let ipv6_dc = config.ipv6.unwrap_or(probe.detected_ipv6.is_some()) && probe.detected_ipv6.is_some();

src/network/stun.rs

@@ -7,6 +7,7 @@ use tokio::net::{lookup_host, UdpSocket};
 use tokio::time::{timeout, Duration, sleep};
 
 use crate::error::{ProxyError, Result};
+use crate::network::dns_overrides::{resolve, split_host_port};
 
 #[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
 pub enum IpFamily {
@@ -40,16 +41,31 @@ pub async fn stun_probe_dual(stun_addr: &str) -> Result<DualStunResult> {
 }
 
 pub async fn stun_probe_family(stun_addr: &str, family: IpFamily) -> Result<Option<StunProbeResult>> {
+    stun_probe_family_with_bind(stun_addr, family, None).await
+}
+
+pub async fn stun_probe_family_with_bind(
+    stun_addr: &str,
+    family: IpFamily,
+    bind_ip: Option<IpAddr>,
+) -> Result<Option<StunProbeResult>> {
     use rand::RngCore;
 
-    let bind_addr = match family {
-        IpFamily::V4 => "0.0.0.0:0",
-        IpFamily::V6 => "[::]:0",
+    let bind_addr = match (family, bind_ip) {
+        (IpFamily::V4, Some(IpAddr::V4(ip))) => SocketAddr::new(IpAddr::V4(ip), 0),
+        (IpFamily::V6, Some(IpAddr::V6(ip))) => SocketAddr::new(IpAddr::V6(ip), 0),
+        (IpFamily::V4, Some(IpAddr::V6(_))) | (IpFamily::V6, Some(IpAddr::V4(_))) => {
+            return Ok(None);
+        }
+        (IpFamily::V4, None) => SocketAddr::new(IpAddr::V4(Ipv4Addr::UNSPECIFIED), 0),
+        (IpFamily::V6, None) => SocketAddr::new(IpAddr::V6(Ipv6Addr::UNSPECIFIED), 0),
     };
 
-    let socket = UdpSocket::bind(bind_addr)
-        .await
-        .map_err(|e| ProxyError::Proxy(format!("STUN bind failed: {e}")))?;
+    let socket = match UdpSocket::bind(bind_addr).await {
+        Ok(socket) => socket,
+        Err(_) if bind_ip.is_some() => return Ok(None),
+        Err(e) => return Err(ProxyError::Proxy(format!("STUN bind failed: {e}"))),
+    };
 
     let target_addr = resolve_stun_addr(stun_addr, family).await?;
     if let Some(addr) = target_addr {
@@ -198,6 +214,16 @@ async fn resolve_stun_addr(stun_addr: &str, family: IpFamily) -> Result<Option<S
         });
     }
 
+    if let Some((host, port)) = split_host_port(stun_addr)
+        && let Some(ip) = resolve(&host, port)
+    {
+        let addr = SocketAddr::new(ip, port);
+        return Ok(match (addr.is_ipv4(), family) {
+            (true, IpFamily::V4) | (false, IpFamily::V6) => Some(addr),
+            _ => None,
+        });
+    }
+
     let mut addrs = lookup_host(stun_addr)
         .await
         .map_err(|e| ProxyError::Proxy(format!("STUN resolve failed: {e}")))?;

src/proxy/client.rs

@@ -1,7 +1,7 @@
 //! Client Handler
 
 use std::future::Future;
-use std::net::SocketAddr;
+use std::net::{IpAddr, SocketAddr};
 use std::pin::Pin;
 use std::sync::Arc;
 use std::time::Duration;
@@ -27,6 +27,7 @@ use crate::error::{HandshakeResult, ProxyError, Result};
 use crate::ip_tracker::UserIpTracker;
 use crate::protocol::constants::*;
 use crate::protocol::tls;
+use crate::stats::beobachten::BeobachtenStore;
 use crate::stats::{ReplayChecker, Stats};
 use crate::stream::{BufferPool, CryptoReader, CryptoWriter};
 use crate::transport::middle_proxy::MePool;
@@ -39,6 +40,36 @@ use crate::proxy::handshake::{HandshakeSuccess, handle_mtproto_handshake, handle
 use crate::proxy::masking::handle_bad_client;
 use crate::proxy::middle_relay::handle_via_middle_proxy;
 
+fn beobachten_ttl(config: &ProxyConfig) -> Duration {
+    Duration::from_secs(config.general.beobachten_minutes.saturating_mul(60))
+}
+
+fn record_beobachten_class(
+    beobachten: &BeobachtenStore,
+    config: &ProxyConfig,
+    peer_ip: IpAddr,
+    class: &str,
+) {
+    if !config.general.beobachten {
+        return;
+    }
+    beobachten.record(class, peer_ip, beobachten_ttl(config));
+}
+
+fn record_handshake_failure_class(
+    beobachten: &BeobachtenStore,
+    config: &ProxyConfig,
+    peer_ip: IpAddr,
+    error: &ProxyError,
+) {
+    let class = if error.to_string().contains("expected 64 bytes, got 0") {
+        "expected_64_got_0"
+    } else {
+        "other"
+    };
+    record_beobachten_class(beobachten, config, peer_ip, class);
+}
+
 pub async fn handle_client_stream<S>(
     mut stream: S,
     peer: SocketAddr,
@@ -51,6 +82,7 @@ pub async fn handle_client_stream<S>(
     me_pool: Option<Arc<MePool>>,
     tls_cache: Option<Arc<TlsFrontCache>>,
     ip_tracker: Arc<UserIpTracker>,
+    beobachten: Arc<BeobachtenStore>,
     proxy_protocol_enabled: bool,
 ) -> Result<()>
 where
@@ -73,6 +105,7 @@ where
             Err(e) => {
                 stats.increment_connects_bad();
                 warn!(peer = %peer, error = %e, "Invalid PROXY protocol header");
+                record_beobachten_class(&beobachten, &config, peer.ip(), "other");
                 return Err(e);
             }
         }
@@ -82,6 +115,9 @@ where
 
     let handshake_timeout = Duration::from_secs(config.timeouts.client_handshake);
     let stats_for_timeout = stats.clone();
+    let config_for_timeout = config.clone();
+    let beobachten_for_timeout = beobachten.clone();
+    let peer_for_timeout = real_peer.ip();
 
     // For non-TCP streams, use a synthetic local address
     let local_addr: SocketAddr = format!("0.0.0.0:{}", config.server.port)
@@ -103,7 +139,15 @@ where
                 debug!(peer = %real_peer, tls_len = tls_len, "TLS handshake too short");
                 stats.increment_connects_bad();
                 let (reader, writer) = tokio::io::split(stream);
-                handle_bad_client(reader, writer, &first_bytes, &config).await;
+                handle_bad_client(
+                    reader,
+                    writer,
+                    &first_bytes,
+                    real_peer,
+                    &config,
+                    &beobachten,
+                )
+                .await;
                 return Ok(HandshakeOutcome::Handled);
             }
 
@@ -120,7 +164,15 @@ where
                 HandshakeResult::Success(result) => result,
                 HandshakeResult::BadClient { reader, writer } => {
                     stats.increment_connects_bad();
-                    handle_bad_client(reader, writer, &handshake, &config).await;
+                    handle_bad_client(
+                        reader,
+                        writer,
+                        &handshake,
+                        real_peer,
+                        &config,
+                        &beobachten,
+                    )
+                    .await;
                     return Ok(HandshakeOutcome::Handled);
                 }
                 HandshakeResult::Error(e) => return Err(e),
@@ -156,7 +208,15 @@ where
                 debug!(peer = %real_peer, "Non-TLS modes disabled");
                 stats.increment_connects_bad();
                 let (reader, writer) = tokio::io::split(stream);
-                handle_bad_client(reader, writer, &first_bytes, &config).await;
+                handle_bad_client(
+                    reader,
+                    writer,
+                    &first_bytes,
+                    real_peer,
+                    &config,
+                    &beobachten,
+                )
+                .await;
                 return Ok(HandshakeOutcome::Handled);
             }
 
@@ -173,7 +233,15 @@ where
                 HandshakeResult::Success(result) => result,
                 HandshakeResult::BadClient { reader, writer } => {
                     stats.increment_connects_bad();
-                    handle_bad_client(reader, writer, &handshake, &config).await;
+                    handle_bad_client(
+                        reader,
+                        writer,
+                        &handshake,
+                        real_peer,
+                        &config,
+                        &beobachten,
+                    )
+                    .await;
                     return Ok(HandshakeOutcome::Handled);
                 }
                 HandshakeResult::Error(e) => return Err(e),
@@ -200,11 +268,23 @@ where
         Ok(Ok(outcome)) => outcome,
         Ok(Err(e)) => {
             debug!(peer = %peer, error = %e, "Handshake failed");
+            record_handshake_failure_class(
+                &beobachten_for_timeout,
+                &config_for_timeout,
+                peer_for_timeout,
+                &e,
+            );
             return Err(e);
         }
         Err(_) => {
             stats_for_timeout.increment_handshake_timeouts();
             debug!(peer = %peer, "Handshake timeout");
+            record_beobachten_class(
+                &beobachten_for_timeout,
+                &config_for_timeout,
+                peer_for_timeout,
+                "other",
+            );
             return Err(ProxyError::TgHandshakeTimeout);
         }
     };
@@ -230,6 +310,7 @@ pub struct RunningClientHandler {
     me_pool: Option<Arc<MePool>>,
     tls_cache: Option<Arc<TlsFrontCache>>,
     ip_tracker: Arc<UserIpTracker>,
+    beobachten: Arc<BeobachtenStore>,
     proxy_protocol_enabled: bool,
 }
 
@@ -246,6 +327,7 @@ impl ClientHandler {
         me_pool: Option<Arc<MePool>>,
         tls_cache: Option<Arc<TlsFrontCache>>,
         ip_tracker: Arc<UserIpTracker>,
+        beobachten: Arc<BeobachtenStore>,
         proxy_protocol_enabled: bool,
     ) -> RunningClientHandler {
         RunningClientHandler {
@@ -260,6 +342,7 @@ impl ClientHandler {
             me_pool,
             tls_cache,
             ip_tracker,
+            beobachten,
             proxy_protocol_enabled,
         }
     }
@@ -284,17 +367,32 @@ impl RunningClientHandler {
 
         let handshake_timeout = Duration::from_secs(self.config.timeouts.client_handshake);
         let stats = self.stats.clone();
+        let config_for_timeout = self.config.clone();
+        let beobachten_for_timeout = self.beobachten.clone();
+        let peer_for_timeout = peer.ip();
 
         // Phase 1: handshake (with timeout)
         let outcome = match timeout(handshake_timeout, self.do_handshake()).await {
             Ok(Ok(outcome)) => outcome,
             Ok(Err(e)) => {
                 debug!(peer = %peer, error = %e, "Handshake failed");
+                record_handshake_failure_class(
+                    &beobachten_for_timeout,
+                    &config_for_timeout,
+                    peer_for_timeout,
+                    &e,
+                );
                 return Err(e);
             }
             Err(_) => {
                 stats.increment_handshake_timeouts();
                 debug!(peer = %peer, "Handshake timeout");
+                record_beobachten_class(
+                    &beobachten_for_timeout,
+                    &config_for_timeout,
+                    peer_for_timeout,
+                    "other",
+                );
                 return Err(ProxyError::TgHandshakeTimeout);
             }
         };
@@ -321,6 +419,12 @@ impl RunningClientHandler {
                 Err(e) => {
                     self.stats.increment_connects_bad();
                     warn!(peer = %self.peer, error = %e, "Invalid PROXY protocol header");
+                    record_beobachten_class(
+                        &self.beobachten,
+                        &self.config,
+                        self.peer.ip(),
+                        "other",
+                    );
                     return Err(e);
                 }
             }
@@ -354,7 +458,15 @@ impl RunningClientHandler {
             debug!(peer = %peer, tls_len = tls_len, "TLS handshake too short");
             self.stats.increment_connects_bad();
             let (reader, writer) = self.stream.into_split();
-            handle_bad_client(reader, writer, &first_bytes, &self.config).await;
+            handle_bad_client(
+                reader,
+                writer,
+                &first_bytes,
+                peer,
+                &self.config,
+                &self.beobachten,
+            )
+            .await;
             return Ok(HandshakeOutcome::Handled);
         }
 
@@ -385,7 +497,15 @@ impl RunningClientHandler {
             HandshakeResult::Success(result) => result,
             HandshakeResult::BadClient { reader, writer } => {
                 stats.increment_connects_bad();
-                handle_bad_client(reader, writer, &handshake, &config).await;
+                handle_bad_client(
+                    reader,
+                    writer,
+                    &handshake,
+                    peer,
+                    &config,
+                    &self.beobachten,
+                )
+                .await;
                 return Ok(HandshakeOutcome::Handled);
             }
             HandshakeResult::Error(e) => return Err(e),
@@ -446,7 +566,15 @@ impl RunningClientHandler {
             debug!(peer = %peer, "Non-TLS modes disabled");
             self.stats.increment_connects_bad();
             let (reader, writer) = self.stream.into_split();
-            handle_bad_client(reader, writer, &first_bytes, &self.config).await;
+            handle_bad_client(
+                reader,
+                writer,
+                &first_bytes,
+                peer,
+                &self.config,
+                &self.beobachten,
+            )
+            .await;
             return Ok(HandshakeOutcome::Handled);
         }
 
@@ -476,7 +604,15 @@ impl RunningClientHandler {
             HandshakeResult::Success(result) => result,
             HandshakeResult::BadClient { reader, writer } => {
                 stats.increment_connects_bad();
-                handle_bad_client(reader, writer, &handshake, &config).await;
+                handle_bad_client(
+                    reader,
+                    writer,
+                    &handshake,
+                    peer,
+                    &config,
+                    &self.beobachten,
+                )
+                .await;
                 return Ok(HandshakeOutcome::Handled);
             }
             HandshakeResult::Error(e) => return Err(e),

src/proxy/masking.rs

@@ -1,6 +1,7 @@
 //! Masking - forward unrecognized traffic to mask host
 
 use std::str;
+use std::net::SocketAddr;
 use std::time::Duration;
 use tokio::net::TcpStream;
 #[cfg(unix)]
@@ -9,6 +10,9 @@ use tokio::io::{AsyncRead, AsyncWrite, AsyncReadExt, AsyncWriteExt};
 use tokio::time::timeout;
 use tracing::debug;
 use crate::config::ProxyConfig;
+use crate::network::dns_overrides::resolve_socket_addr;
+use crate::stats::beobachten::BeobachtenStore;
+use crate::transport::proxy_protocol::{ProxyProtocolV1Builder, ProxyProtocolV2Builder};
 
 const MASK_TIMEOUT: Duration = Duration::from_secs(5);
 /// Maximum duration for the entire masking relay.
@@ -50,20 +54,26 @@ pub async fn handle_bad_client<R, W>(
     reader: R,
     writer: W,
     initial_data: &[u8],
+    peer: SocketAddr,
     config: &ProxyConfig,
+    beobachten: &BeobachtenStore,
 )
 where
     R: AsyncRead + Unpin + Send + 'static,
     W: AsyncWrite + Unpin + Send + 'static,
 {
+    let client_type = detect_client_type(initial_data);
+    if config.general.beobachten {
+        let ttl = Duration::from_secs(config.general.beobachten_minutes.saturating_mul(60));
+        beobachten.record(client_type, peer.ip(), ttl);
+    }
+
     if !config.censorship.mask {
         // Masking disabled, just consume data
         consume_client_data(reader).await;
         return;
     }
 
-    let client_type = detect_client_type(initial_data);
-
     // Connect via Unix socket or TCP
     #[cfg(unix)]
     if let Some(ref sock_path) = config.censorship.mask_unix_sock {
@@ -106,12 +116,44 @@ where
         "Forwarding bad client to mask host"
     );
 
-    // Connect to mask host
-    let mask_addr = format!("{}:{}", mask_host, mask_port);
+    // Apply runtime DNS override for mask target when configured.
+    let mask_addr = resolve_socket_addr(mask_host, mask_port)
+        .map(|addr| addr.to_string())
+        .unwrap_or_else(|| format!("{}:{}", mask_host, mask_port));
     let connect_result = timeout(MASK_TIMEOUT, TcpStream::connect(&mask_addr)).await;
     match connect_result {
         Ok(Ok(stream)) => {
-            let (mask_read, mask_write) = stream.into_split();
+            let proxy_header: Option<Vec<u8>> = match config.censorship.mask_proxy_protocol {
+                0 => None,
+                version => {
+                    let header = if let Ok(local_addr) = stream.local_addr() {
+                        match version {
+                            2 => ProxyProtocolV2Builder::new().with_addrs(peer, local_addr).build(),
+                            _ => match (peer, local_addr) {
+                                (SocketAddr::V4(src), SocketAddr::V4(dst)) =>
+                                    ProxyProtocolV1Builder::new().tcp4(src.into(), dst.into()).build(),
+                                (SocketAddr::V6(src), SocketAddr::V6(dst)) =>
+                                    ProxyProtocolV1Builder::new().tcp6(src.into(), dst.into()).build(),
+                                _ =>
+                                    ProxyProtocolV1Builder::new().build(),
+                            },
+                        }
+                    } else {
+                        match version {
+                            2 => ProxyProtocolV2Builder::new().build(),
+                            _ => ProxyProtocolV1Builder::new().build(),
+                        }
+                    };
+                    Some(header)
+                }
+            };
+
+            let (mask_read, mut mask_write) = stream.into_split();
+            if let Some(header) = proxy_header {
+                if mask_write.write_all(&header).await.is_err() {
+                    return;
+                }
+            }
             if timeout(MASK_RELAY_TIMEOUT, relay_to_mask(reader, writer, mask_read, mask_write, initial_data)).await.is_err() {
                 debug!("Mask relay timed out");
             }

src/stats/beobachten.rs

@@ -0,0 +1,117 @@
+//! Per-IP forensic buckets for scanner and handshake failure observation.
+
+use std::collections::{BTreeMap, HashMap};
+use std::net::IpAddr;
+use std::time::{Duration, Instant};
+
+use parking_lot::Mutex;
+
+const CLEANUP_INTERVAL: Duration = Duration::from_secs(30);
+
+#[derive(Default)]
+struct BeobachtenInner {
+    entries: HashMap<(String, IpAddr), BeobachtenEntry>,
+    last_cleanup: Option<Instant>,
+}
+
+#[derive(Clone, Copy)]
+struct BeobachtenEntry {
+    tries: u64,
+    last_seen: Instant,
+}
+
+/// In-memory, TTL-scoped per-IP counters keyed by source class.
+pub struct BeobachtenStore {
+    inner: Mutex<BeobachtenInner>,
+}
+
+impl Default for BeobachtenStore {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
+impl BeobachtenStore {
+    pub fn new() -> Self {
+        Self {
+            inner: Mutex::new(BeobachtenInner::default()),
+        }
+    }
+
+    pub fn record(&self, class: &str, ip: IpAddr, ttl: Duration) {
+        if class.is_empty() || ttl.is_zero() {
+            return;
+        }
+
+        let now = Instant::now();
+        let mut guard = self.inner.lock();
+        Self::cleanup_if_needed(&mut guard, now, ttl);
+
+        let key = (class.to_string(), ip);
+        let entry = guard.entries.entry(key).or_insert(BeobachtenEntry {
+            tries: 0,
+            last_seen: now,
+        });
+        entry.tries = entry.tries.saturating_add(1);
+        entry.last_seen = now;
+    }
+
+    pub fn snapshot_text(&self, ttl: Duration) -> String {
+        if ttl.is_zero() {
+            return "beobachten disabled\n".to_string();
+        }
+
+        let now = Instant::now();
+        let mut guard = self.inner.lock();
+        Self::cleanup(&mut guard, now, ttl);
+        guard.last_cleanup = Some(now);
+
+        let mut grouped = BTreeMap::<String, Vec<(IpAddr, u64)>>::new();
+        for ((class, ip), entry) in &guard.entries {
+            grouped
+                .entry(class.clone())
+                .or_default()
+                .push((*ip, entry.tries));
+        }
+
+        if grouped.is_empty() {
+            return "empty\n".to_string();
+        }
+
+        let mut out = String::with_capacity(grouped.len() * 64);
+        for (class, entries) in &mut grouped {
+            out.push('[');
+            out.push_str(class);
+            out.push_str("]\n");
+
+            entries.sort_by(|(ip_a, tries_a), (ip_b, tries_b)| {
+                tries_b
+                    .cmp(tries_a)
+                    .then_with(|| ip_a.to_string().cmp(&ip_b.to_string()))
+            });
+
+            for (ip, tries) in entries {
+                out.push_str(&format!("{ip}-{tries}\n"));
+            }
+        }
+
+        out
+    }
+
+    fn cleanup_if_needed(inner: &mut BeobachtenInner, now: Instant, ttl: Duration) {
+        let should_cleanup = match inner.last_cleanup {
+            Some(last) => now.saturating_duration_since(last) >= CLEANUP_INTERVAL,
+            None => true,
+        };
+        if should_cleanup {
+            Self::cleanup(inner, now, ttl);
+            inner.last_cleanup = Some(now);
+        }
+    }
+
+    fn cleanup(inner: &mut BeobachtenInner, now: Instant, ttl: Duration) {
+        inner.entries.retain(|_, entry| {
+            now.saturating_duration_since(entry.last_seen) <= ttl
+        });
+    }
+}

src/stats/mod.rs

@@ -2,7 +2,10 @@
 
 #![allow(dead_code)]
 
-use std::sync::atomic::{AtomicU64, Ordering};
+pub mod beobachten;
+pub mod telemetry;
+
+use std::sync::atomic::{AtomicBool, AtomicU8, AtomicU64, Ordering};
 use std::time::{Instant, Duration};
 use dashmap::DashMap;
 use parking_lot::Mutex;
@@ -13,6 +16,9 @@ use std::collections::hash_map::DefaultHasher;
 use std::collections::VecDeque;
 use tracing::debug;
 
+use crate::config::MeTelemetryLevel;
+use self::telemetry::TelemetryPolicy;
+
 // ============= Stats =============
 
 #[derive(Default)]
@@ -31,6 +37,10 @@ pub struct Stats {
     me_route_drop_no_conn: AtomicU64,
     me_route_drop_channel_closed: AtomicU64,
     me_route_drop_queue_full: AtomicU64,
+    me_route_drop_queue_full_base: AtomicU64,
+    me_route_drop_queue_full_high: AtomicU64,
+    me_socks_kdf_strict_reject: AtomicU64,
+    me_socks_kdf_compat_fallback: AtomicU64,
     secure_padding_invalid: AtomicU64,
     desync_total: AtomicU64,
     desync_full_logged: AtomicU64,
@@ -43,6 +53,16 @@ pub struct Stats {
     pool_drain_active: AtomicU64,
     pool_force_close_total: AtomicU64,
     pool_stale_pick_total: AtomicU64,
+    me_writer_removed_total: AtomicU64,
+    me_writer_removed_unexpected_total: AtomicU64,
+    me_refill_triggered_total: AtomicU64,
+    me_refill_skipped_inflight_total: AtomicU64,
+    me_refill_failed_total: AtomicU64,
+    me_writer_restored_same_endpoint_total: AtomicU64,
+    me_writer_restored_fallback_total: AtomicU64,
+    telemetry_core_enabled: AtomicBool,
+    telemetry_user_enabled: AtomicBool,
+    telemetry_me_level: AtomicU8,
     user_stats: DashMap<String, UserStats>,
     start_time: parking_lot::RwLock<Option<Instant>>,
 }
@@ -60,44 +80,167 @@ pub struct UserStats {
 impl Stats {
     pub fn new() -> Self {
         let stats = Self::default();
+        stats.apply_telemetry_policy(TelemetryPolicy::default());
         *stats.start_time.write() = Some(Instant::now());
         stats
     }
-    
-    pub fn increment_connects_all(&self) { self.connects_all.fetch_add(1, Ordering::Relaxed); }
-    pub fn increment_connects_bad(&self) { self.connects_bad.fetch_add(1, Ordering::Relaxed); }
-    pub fn increment_handshake_timeouts(&self) { self.handshake_timeouts.fetch_add(1, Ordering::Relaxed); }
-    pub fn increment_me_keepalive_sent(&self) { self.me_keepalive_sent.fetch_add(1, Ordering::Relaxed); }
-    pub fn increment_me_keepalive_failed(&self) { self.me_keepalive_failed.fetch_add(1, Ordering::Relaxed); }
-    pub fn increment_me_keepalive_pong(&self) { self.me_keepalive_pong.fetch_add(1, Ordering::Relaxed); }
-    pub fn increment_me_keepalive_timeout(&self) { self.me_keepalive_timeout.fetch_add(1, Ordering::Relaxed); }
-    pub fn increment_me_keepalive_timeout_by(&self, value: u64) {
-        self.me_keepalive_timeout.fetch_add(value, Ordering::Relaxed);
+
+    fn telemetry_me_level(&self) -> MeTelemetryLevel {
+        MeTelemetryLevel::from_u8(self.telemetry_me_level.load(Ordering::Relaxed))
+    }
+
+    fn telemetry_core_enabled(&self) -> bool {
+        self.telemetry_core_enabled.load(Ordering::Relaxed)
+    }
+
+    fn telemetry_user_enabled(&self) -> bool {
+        self.telemetry_user_enabled.load(Ordering::Relaxed)
+    }
+
+    fn telemetry_me_allows_normal(&self) -> bool {
+        self.telemetry_me_level().allows_normal()
+    }
+
+    fn telemetry_me_allows_debug(&self) -> bool {
+        self.telemetry_me_level().allows_debug()
+    }
+
+    pub fn apply_telemetry_policy(&self, policy: TelemetryPolicy) {
+        self.telemetry_core_enabled
+            .store(policy.core_enabled, Ordering::Relaxed);
+        self.telemetry_user_enabled
+            .store(policy.user_enabled, Ordering::Relaxed);
+        self.telemetry_me_level
+            .store(policy.me_level.as_u8(), Ordering::Relaxed);
+    }
+
+    pub fn telemetry_policy(&self) -> TelemetryPolicy {
+        TelemetryPolicy {
+            core_enabled: self.telemetry_core_enabled(),
+            user_enabled: self.telemetry_user_enabled(),
+            me_level: self.telemetry_me_level(),
+        }
+    }
+    
+    pub fn increment_connects_all(&self) {
+        if self.telemetry_core_enabled() {
+            self.connects_all.fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_connects_bad(&self) {
+        if self.telemetry_core_enabled() {
+            self.connects_bad.fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_handshake_timeouts(&self) {
+        if self.telemetry_core_enabled() {
+            self.handshake_timeouts.fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_keepalive_sent(&self) {
+        if self.telemetry_me_allows_debug() {
+            self.me_keepalive_sent.fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_keepalive_failed(&self) {
+        if self.telemetry_me_allows_normal() {
+            self.me_keepalive_failed.fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_keepalive_pong(&self) {
+        if self.telemetry_me_allows_debug() {
+            self.me_keepalive_pong.fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_keepalive_timeout(&self) {
+        if self.telemetry_me_allows_normal() {
+            self.me_keepalive_timeout.fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_keepalive_timeout_by(&self, value: u64) {
+        if self.telemetry_me_allows_normal() {
+            self.me_keepalive_timeout.fetch_add(value, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_reconnect_attempt(&self) {
+        if self.telemetry_me_allows_normal() {
+            self.me_reconnect_attempts.fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_reconnect_success(&self) {
+        if self.telemetry_me_allows_normal() {
+            self.me_reconnect_success.fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_crc_mismatch(&self) {
+        if self.telemetry_me_allows_normal() {
+            self.me_crc_mismatch.fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_seq_mismatch(&self) {
+        if self.telemetry_me_allows_normal() {
+            self.me_seq_mismatch.fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_route_drop_no_conn(&self) {
+        if self.telemetry_me_allows_normal() {
+            self.me_route_drop_no_conn.fetch_add(1, Ordering::Relaxed);
+        }
     }
-    pub fn increment_me_reconnect_attempt(&self) { self.me_reconnect_attempts.fetch_add(1, Ordering::Relaxed); }
-    pub fn increment_me_reconnect_success(&self) { self.me_reconnect_success.fetch_add(1, Ordering::Relaxed); }
-    pub fn increment_me_crc_mismatch(&self) { self.me_crc_mismatch.fetch_add(1, Ordering::Relaxed); }
-    pub fn increment_me_seq_mismatch(&self) { self.me_seq_mismatch.fetch_add(1, Ordering::Relaxed); }
-    pub fn increment_me_route_drop_no_conn(&self) { self.me_route_drop_no_conn.fetch_add(1, Ordering::Relaxed); }
     pub fn increment_me_route_drop_channel_closed(&self) {
-        self.me_route_drop_channel_closed.fetch_add(1, Ordering::Relaxed);
+        if self.telemetry_me_allows_normal() {
+            self.me_route_drop_channel_closed.fetch_add(1, Ordering::Relaxed);
+        }
     }
     pub fn increment_me_route_drop_queue_full(&self) {
-        self.me_route_drop_queue_full.fetch_add(1, Ordering::Relaxed);
+        if self.telemetry_me_allows_normal() {
+            self.me_route_drop_queue_full.fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_route_drop_queue_full_base(&self) {
+        if self.telemetry_me_allows_normal() {
+            self.me_route_drop_queue_full_base.fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_route_drop_queue_full_high(&self) {
+        if self.telemetry_me_allows_normal() {
+            self.me_route_drop_queue_full_high.fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_socks_kdf_strict_reject(&self) {
+        if self.telemetry_me_allows_normal() {
+            self.me_socks_kdf_strict_reject.fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_socks_kdf_compat_fallback(&self) {
+        if self.telemetry_me_allows_debug() {
+            self.me_socks_kdf_compat_fallback.fetch_add(1, Ordering::Relaxed);
+        }
     }
     pub fn increment_secure_padding_invalid(&self) {
-        self.secure_padding_invalid.fetch_add(1, Ordering::Relaxed);
+        if self.telemetry_me_allows_normal() {
+            self.secure_padding_invalid.fetch_add(1, Ordering::Relaxed);
+        }
     }
     pub fn increment_desync_total(&self) {
-        self.desync_total.fetch_add(1, Ordering::Relaxed);
+        if self.telemetry_me_allows_normal() {
+            self.desync_total.fetch_add(1, Ordering::Relaxed);
+        }
     }
     pub fn increment_desync_full_logged(&self) {
-        self.desync_full_logged.fetch_add(1, Ordering::Relaxed);
+        if self.telemetry_me_allows_normal() {
+            self.desync_full_logged.fetch_add(1, Ordering::Relaxed);
+        }
     }
     pub fn increment_desync_suppressed(&self) {
-        self.desync_suppressed.fetch_add(1, Ordering::Relaxed);
+        if self.telemetry_me_allows_normal() {
+            self.desync_suppressed.fetch_add(1, Ordering::Relaxed);
+        }
     }
     pub fn observe_desync_frames_ok(&self, frames_ok: u64) {
+        if !self.telemetry_me_allows_normal() {
+            return;
+        }
         match frames_ok {
             0 => {
                 self.desync_frames_bucket_0.fetch_add(1, Ordering::Relaxed);
@@ -114,12 +257,19 @@ impl Stats {
         }
     }
     pub fn increment_pool_swap_total(&self) {
-        self.pool_swap_total.fetch_add(1, Ordering::Relaxed);
+        if self.telemetry_me_allows_debug() {
+            self.pool_swap_total.fetch_add(1, Ordering::Relaxed);
+        }
     }
     pub fn increment_pool_drain_active(&self) {
-        self.pool_drain_active.fetch_add(1, Ordering::Relaxed);
+        if self.telemetry_me_allows_debug() {
+            self.pool_drain_active.fetch_add(1, Ordering::Relaxed);
+        }
     }
     pub fn decrement_pool_drain_active(&self) {
+        if !self.telemetry_me_allows_debug() {
+            return;
+        }
         let mut current = self.pool_drain_active.load(Ordering::Relaxed);
         loop {
             if current == 0 {
@@ -137,10 +287,51 @@ impl Stats {
         }
     }
     pub fn increment_pool_force_close_total(&self) {
-        self.pool_force_close_total.fetch_add(1, Ordering::Relaxed);
+        if self.telemetry_me_allows_normal() {
+            self.pool_force_close_total.fetch_add(1, Ordering::Relaxed);
+        }
     }
     pub fn increment_pool_stale_pick_total(&self) {
-        self.pool_stale_pick_total.fetch_add(1, Ordering::Relaxed);
+        if self.telemetry_me_allows_normal() {
+            self.pool_stale_pick_total.fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_writer_removed_total(&self) {
+        if self.telemetry_me_allows_debug() {
+            self.me_writer_removed_total.fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_writer_removed_unexpected_total(&self) {
+        if self.telemetry_me_allows_normal() {
+            self.me_writer_removed_unexpected_total.fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_refill_triggered_total(&self) {
+        if self.telemetry_me_allows_debug() {
+            self.me_refill_triggered_total.fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_refill_skipped_inflight_total(&self) {
+        if self.telemetry_me_allows_debug() {
+            self.me_refill_skipped_inflight_total.fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_refill_failed_total(&self) {
+        if self.telemetry_me_allows_normal() {
+            self.me_refill_failed_total.fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_writer_restored_same_endpoint_total(&self) {
+        if self.telemetry_me_allows_normal() {
+            self.me_writer_restored_same_endpoint_total
+                .fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_writer_restored_fallback_total(&self) {
+        if self.telemetry_me_allows_normal() {
+            self.me_writer_restored_fallback_total
+                .fetch_add(1, Ordering::Relaxed);
+        }
     }
     pub fn get_connects_all(&self) -> u64 { self.connects_all.load(Ordering::Relaxed) }
     pub fn get_connects_bad(&self) -> u64 { self.connects_bad.load(Ordering::Relaxed) }
@@ -159,6 +350,18 @@ impl Stats {
     pub fn get_me_route_drop_queue_full(&self) -> u64 {
         self.me_route_drop_queue_full.load(Ordering::Relaxed)
     }
+    pub fn get_me_route_drop_queue_full_base(&self) -> u64 {
+        self.me_route_drop_queue_full_base.load(Ordering::Relaxed)
+    }
+    pub fn get_me_route_drop_queue_full_high(&self) -> u64 {
+        self.me_route_drop_queue_full_high.load(Ordering::Relaxed)
+    }
+    pub fn get_me_socks_kdf_strict_reject(&self) -> u64 {
+        self.me_socks_kdf_strict_reject.load(Ordering::Relaxed)
+    }
+    pub fn get_me_socks_kdf_compat_fallback(&self) -> u64 {
+        self.me_socks_kdf_compat_fallback.load(Ordering::Relaxed)
+    }
     pub fn get_secure_padding_invalid(&self) -> u64 {
         self.secure_padding_invalid.load(Ordering::Relaxed)
     }
@@ -195,13 +398,40 @@ impl Stats {
     pub fn get_pool_stale_pick_total(&self) -> u64 {
         self.pool_stale_pick_total.load(Ordering::Relaxed)
     }
+    pub fn get_me_writer_removed_total(&self) -> u64 {
+        self.me_writer_removed_total.load(Ordering::Relaxed)
+    }
+    pub fn get_me_writer_removed_unexpected_total(&self) -> u64 {
+        self.me_writer_removed_unexpected_total.load(Ordering::Relaxed)
+    }
+    pub fn get_me_refill_triggered_total(&self) -> u64 {
+        self.me_refill_triggered_total.load(Ordering::Relaxed)
+    }
+    pub fn get_me_refill_skipped_inflight_total(&self) -> u64 {
+        self.me_refill_skipped_inflight_total.load(Ordering::Relaxed)
+    }
+    pub fn get_me_refill_failed_total(&self) -> u64 {
+        self.me_refill_failed_total.load(Ordering::Relaxed)
+    }
+    pub fn get_me_writer_restored_same_endpoint_total(&self) -> u64 {
+        self.me_writer_restored_same_endpoint_total.load(Ordering::Relaxed)
+    }
+    pub fn get_me_writer_restored_fallback_total(&self) -> u64 {
+        self.me_writer_restored_fallback_total.load(Ordering::Relaxed)
+    }
     
     pub fn increment_user_connects(&self, user: &str) {
+        if !self.telemetry_user_enabled() {
+            return;
+        }
         self.user_stats.entry(user.to_string()).or_default()
             .connects.fetch_add(1, Ordering::Relaxed);
     }
     
     pub fn increment_user_curr_connects(&self, user: &str) {
+        if !self.telemetry_user_enabled() {
+            return;
+        }
         self.user_stats.entry(user.to_string()).or_default()
             .curr_connects.fetch_add(1, Ordering::Relaxed);
     }
@@ -234,21 +464,33 @@ impl Stats {
     }
     
     pub fn add_user_octets_from(&self, user: &str, bytes: u64) {
+        if !self.telemetry_user_enabled() {
+            return;
+        }
         self.user_stats.entry(user.to_string()).or_default()
             .octets_from_client.fetch_add(bytes, Ordering::Relaxed);
     }
     
     pub fn add_user_octets_to(&self, user: &str, bytes: u64) {
+        if !self.telemetry_user_enabled() {
+            return;
+        }
         self.user_stats.entry(user.to_string()).or_default()
             .octets_to_client.fetch_add(bytes, Ordering::Relaxed);
     }
     
     pub fn increment_user_msgs_from(&self, user: &str) {
+        if !self.telemetry_user_enabled() {
+            return;
+        }
         self.user_stats.entry(user.to_string()).or_default()
             .msgs_from_client.fetch_add(1, Ordering::Relaxed);
     }
     
     pub fn increment_user_msgs_to(&self, user: &str) {
+        if !self.telemetry_user_enabled() {
+            return;
+        }
         self.user_stats.entry(user.to_string()).or_default()
             .msgs_to_client.fetch_add(1, Ordering::Relaxed);
     }
@@ -497,6 +739,7 @@ impl ReplayStats {
 #[cfg(test)]
 mod tests {
     use super::*;
+    use crate::config::MeTelemetryLevel;
     use std::sync::Arc;
     
     #[test]
@@ -507,6 +750,40 @@ mod tests {
         stats.increment_connects_all();
         assert_eq!(stats.get_connects_all(), 3);
     }
+
+    #[test]
+    fn test_telemetry_policy_disables_core_and_user_counters() {
+        let stats = Stats::new();
+        stats.apply_telemetry_policy(TelemetryPolicy {
+            core_enabled: false,
+            user_enabled: false,
+            me_level: MeTelemetryLevel::Normal,
+        });
+
+        stats.increment_connects_all();
+        stats.increment_user_connects("alice");
+        stats.add_user_octets_from("alice", 1024);
+        assert_eq!(stats.get_connects_all(), 0);
+        assert_eq!(stats.get_user_curr_connects("alice"), 0);
+        assert_eq!(stats.get_user_total_octets("alice"), 0);
+    }
+
+    #[test]
+    fn test_telemetry_policy_me_silent_blocks_me_counters() {
+        let stats = Stats::new();
+        stats.apply_telemetry_policy(TelemetryPolicy {
+            core_enabled: true,
+            user_enabled: true,
+            me_level: MeTelemetryLevel::Silent,
+        });
+
+        stats.increment_me_crc_mismatch();
+        stats.increment_me_keepalive_sent();
+        stats.increment_me_route_drop_queue_full();
+        assert_eq!(stats.get_me_crc_mismatch(), 0);
+        assert_eq!(stats.get_me_keepalive_sent(), 0);
+        assert_eq!(stats.get_me_route_drop_queue_full(), 0);
+    }
     
     #[test]
     fn test_replay_checker_basic() {

src/stats/telemetry.rs

@@ -0,0 +1,29 @@
+use crate::config::{MeTelemetryLevel, TelemetryConfig};
+
+/// Runtime telemetry policy used by hot-path counters.
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub struct TelemetryPolicy {
+    pub core_enabled: bool,
+    pub user_enabled: bool,
+    pub me_level: MeTelemetryLevel,
+}
+
+impl Default for TelemetryPolicy {
+    fn default() -> Self {
+        Self {
+            core_enabled: true,
+            user_enabled: true,
+            me_level: MeTelemetryLevel::Normal,
+        }
+    }
+}
+
+impl TelemetryPolicy {
+    pub fn from_config(cfg: &TelemetryConfig) -> Self {
+        Self {
+            core_enabled: cfg.core_enabled,
+            user_enabled: cfg.user_enabled,
+            me_level: cfg.me_level,
+        }
+    }
+}

src/tls_front/fetcher.rs

@@ -2,8 +2,10 @@ use std::sync::Arc;
 use std::time::Duration;
 
 use anyhow::{Result, anyhow};
-use tokio::io::{AsyncReadExt, AsyncWriteExt};
+use tokio::io::{AsyncRead, AsyncReadExt, AsyncWrite, AsyncWriteExt};
 use tokio::net::TcpStream;
+#[cfg(unix)]
+use tokio::net::UnixStream;
 use tokio::time::timeout;
 use tokio_rustls::client::TlsStream;
 use tokio_rustls::TlsConnector;
@@ -18,7 +20,9 @@ use x509_parser::prelude::FromDer;
 use x509_parser::certificate::X509Certificate;
 
 use crate::crypto::SecureRandom;
+use crate::network::dns_overrides::resolve_socket_addr;
 use crate::protocol::constants::{TLS_RECORD_APPLICATION, TLS_RECORD_HANDSHAKE};
+use crate::transport::proxy_protocol::{ProxyProtocolV1Builder, ProxyProtocolV2Builder};
 use crate::tls_front::types::{
     ParsedCertificateInfo,
     ParsedServerHello,
@@ -210,7 +214,10 @@ fn gen_key_share(rng: &SecureRandom) -> [u8; 32] {
     key
 }
 
-async fn read_tls_record(stream: &mut TcpStream) -> Result<(u8, Vec<u8>)> {
+async fn read_tls_record<S>(stream: &mut S) -> Result<(u8, Vec<u8>)>
+where
+    S: AsyncRead + Unpin,
+{
     let mut header = [0u8; 5];
     stream.read_exact(&mut header).await?;
     let len = u16::from_be_bytes([header[3], header[4]]) as usize;
@@ -332,6 +339,55 @@ fn u24_bytes(value: usize) -> Option<[u8; 3]> {
     ])
 }
 
+async fn connect_with_dns_override(
+    host: &str,
+    port: u16,
+    connect_timeout: Duration,
+) -> Result<TcpStream> {
+    if let Some(addr) = resolve_socket_addr(host, port) {
+        return Ok(timeout(connect_timeout, TcpStream::connect(addr)).await??);
+    }
+    Ok(timeout(connect_timeout, TcpStream::connect((host, port))).await??)
+}
+
+async fn connect_tcp_with_upstream(
+    host: &str,
+    port: u16,
+    connect_timeout: Duration,
+    upstream: Option<std::sync::Arc<crate::transport::UpstreamManager>>,
+) -> Result<TcpStream> {
+    if let Some(manager) = upstream {
+        if let Some(addr) = resolve_socket_addr(host, port) {
+            match manager.connect(addr, None, None).await {
+                Ok(stream) => return Ok(stream),
+                Err(e) => {
+                    warn!(
+                        host = %host,
+                        port = port,
+                        error = %e,
+                        "Upstream connect failed, using direct connect"
+                    );
+                }
+            }
+        } else if let Ok(mut addrs) = tokio::net::lookup_host((host, port)).await {
+            if let Some(addr) = addrs.find(|a| a.is_ipv4()) {
+                match manager.connect(addr, None, None).await {
+                    Ok(stream) => return Ok(stream),
+                    Err(e) => {
+                        warn!(
+                            host = %host,
+                            port = port,
+                            error = %e,
+                            "Upstream connect failed, using direct connect"
+                        );
+                    }
+                }
+            }
+        }
+    }
+    connect_with_dns_override(host, port, connect_timeout).await
+}
+
 fn encode_tls13_certificate_message(cert_chain_der: &[Vec<u8>]) -> Option<Vec<u8>> {
     if cert_chain_der.is_empty() {
         return None;
@@ -361,18 +417,25 @@ fn encode_tls13_certificate_message(cert_chain_der: &[Vec<u8>]) -> Option<Vec<u8
     Some(message)
 }
 
-async fn fetch_via_raw_tls(
-    host: &str,
-    port: u16,
+async fn fetch_via_raw_tls_stream<S>(
+    mut stream: S,
     sni: &str,
     connect_timeout: Duration,
-) -> Result<TlsFetchResult> {
-    let addr = format!("{host}:{port}");
-    let mut stream = timeout(connect_timeout, TcpStream::connect(addr)).await??;
-
+    proxy_protocol: u8,
+) -> Result<TlsFetchResult>
+where
+    S: AsyncRead + AsyncWrite + Unpin,
+{
     let rng = SecureRandom::new();
     let client_hello = build_client_hello(sni, &rng);
     timeout(connect_timeout, async {
+        if proxy_protocol > 0 {
+            let header = match proxy_protocol {
+                2 => ProxyProtocolV2Builder::new().build(),
+                _ => ProxyProtocolV1Builder::new().build(),
+            };
+            stream.write_all(&header).await?;
+        }
         stream.write_all(&client_hello).await?;
         stream.flush().await?;
         Ok::<(), std::io::Error>(())
@@ -418,34 +481,69 @@ async fn fetch_via_raw_tls(
     })
 }
 
-async fn fetch_via_rustls(
+async fn fetch_via_raw_tls(
     host: &str,
     port: u16,
     sni: &str,
     connect_timeout: Duration,
     upstream: Option<std::sync::Arc<crate::transport::UpstreamManager>>,
+    proxy_protocol: u8,
+    unix_sock: Option<&str>,
 ) -> Result<TlsFetchResult> {
-    // rustls handshake path for certificate and basic negotiated metadata.
-    let stream = if let Some(manager) = upstream {
-        // Resolve host to SocketAddr
-        if let Ok(mut addrs) = tokio::net::lookup_host((host, port)).await {
-            if let Some(addr) = addrs.find(|a| a.is_ipv4()) {
-                match manager.connect(addr, None, None).await {
-                    Ok(s) => s,
-                    Err(e) => {
-                        warn!(sni = %sni, error = %e, "Upstream connect failed, using direct connect");
-                        timeout(connect_timeout, TcpStream::connect((host, port))).await??
-                    }
-                }
-            } else {
-                timeout(connect_timeout, TcpStream::connect((host, port))).await??
+    #[cfg(unix)]
+    if let Some(sock_path) = unix_sock {
+        match timeout(connect_timeout, UnixStream::connect(sock_path)).await {
+            Ok(Ok(stream)) => {
+                debug!(
+                    sni = %sni,
+                    sock = %sock_path,
+                    "Raw TLS fetch using mask unix socket"
+                );
+                return fetch_via_raw_tls_stream(stream, sni, connect_timeout, 0).await;
+            }
+            Ok(Err(e)) => {
+                warn!(
+                    sni = %sni,
+                    sock = %sock_path,
+                    error = %e,
+                    "Raw TLS unix socket connect failed, falling back to TCP"
+                );
+            }
+            Err(_) => {
+                warn!(
+                    sni = %sni,
+                    sock = %sock_path,
+                    "Raw TLS unix socket connect timed out, falling back to TCP"
+                );
             }
-        } else {
-            timeout(connect_timeout, TcpStream::connect((host, port))).await??
         }
-    } else {
-        timeout(connect_timeout, TcpStream::connect((host, port))).await??
-    };
+    }
+
+    #[cfg(not(unix))]
+    let _ = unix_sock;
+
+    let stream = connect_tcp_with_upstream(host, port, connect_timeout, upstream).await?;
+    fetch_via_raw_tls_stream(stream, sni, connect_timeout, proxy_protocol).await
+}
+
+async fn fetch_via_rustls_stream<S>(
+    mut stream: S,
+    host: &str,
+    sni: &str,
+    proxy_protocol: u8,
+) -> Result<TlsFetchResult>
+where
+    S: AsyncRead + AsyncWrite + Unpin,
+{
+    // rustls handshake path for certificate and basic negotiated metadata.
+    if proxy_protocol > 0 {
+        let header = match proxy_protocol {
+            2 => ProxyProtocolV2Builder::new().build(),
+            _ => ProxyProtocolV1Builder::new().build(),
+        };
+        stream.write_all(&header).await?;
+        stream.flush().await?;
+    }
 
     let config = build_client_config();
     let connector = TlsConnector::from(config);
@@ -454,7 +552,7 @@ async fn fetch_via_rustls(
         .or_else(|_| ServerName::try_from(host.to_owned()))
         .map_err(|_| RustlsError::General("invalid SNI".into()))?;
 
-    let tls_stream: TlsStream<TcpStream> = connector.connect(server_name, stream).await?;
+    let tls_stream: TlsStream<S> = connector.connect(server_name, stream).await?;
 
     // Extract negotiated parameters and certificates
     let (_io, session) = tls_stream.get_ref();
@@ -515,6 +613,51 @@ async fn fetch_via_rustls(
     })
 }
 
+async fn fetch_via_rustls(
+    host: &str,
+    port: u16,
+    sni: &str,
+    connect_timeout: Duration,
+    upstream: Option<std::sync::Arc<crate::transport::UpstreamManager>>,
+    proxy_protocol: u8,
+    unix_sock: Option<&str>,
+) -> Result<TlsFetchResult> {
+    #[cfg(unix)]
+    if let Some(sock_path) = unix_sock {
+        match timeout(connect_timeout, UnixStream::connect(sock_path)).await {
+            Ok(Ok(stream)) => {
+                debug!(
+                    sni = %sni,
+                    sock = %sock_path,
+                    "Rustls fetch using mask unix socket"
+                );
+                return fetch_via_rustls_stream(stream, host, sni, 0).await;
+            }
+            Ok(Err(e)) => {
+                warn!(
+                    sni = %sni,
+                    sock = %sock_path,
+                    error = %e,
+                    "Rustls unix socket connect failed, falling back to TCP"
+                );
+            }
+            Err(_) => {
+                warn!(
+                    sni = %sni,
+                    sock = %sock_path,
+                    "Rustls unix socket connect timed out, falling back to TCP"
+                );
+            }
+        }
+    }
+
+    #[cfg(not(unix))]
+    let _ = unix_sock;
+
+    let stream = connect_tcp_with_upstream(host, port, connect_timeout, upstream).await?;
+    fetch_via_rustls_stream(stream, host, sni, proxy_protocol).await
+}
+
 /// Fetch real TLS metadata for the given SNI.
 ///
 /// Strategy:
@@ -527,8 +670,20 @@ pub async fn fetch_real_tls(
     sni: &str,
     connect_timeout: Duration,
     upstream: Option<std::sync::Arc<crate::transport::UpstreamManager>>,
+    proxy_protocol: u8,
+    unix_sock: Option<&str>,
 ) -> Result<TlsFetchResult> {
-    let raw_result = match fetch_via_raw_tls(host, port, sni, connect_timeout).await {
+    let raw_result = match fetch_via_raw_tls(
+        host,
+        port,
+        sni,
+        connect_timeout,
+        upstream.clone(),
+        proxy_protocol,
+        unix_sock,
+    )
+    .await
+    {
         Ok(res) => Some(res),
         Err(e) => {
             warn!(sni = %sni, error = %e, "Raw TLS fetch failed");
@@ -536,7 +691,17 @@ pub async fn fetch_real_tls(
         }
     };
 
-    match fetch_via_rustls(host, port, sni, connect_timeout, upstream).await {
+    match fetch_via_rustls(
+        host,
+        port,
+        sni,
+        connect_timeout,
+        upstream,
+        proxy_protocol,
+        unix_sock,
+    )
+    .await
+    {
         Ok(rustls_result) => {
             if let Some(mut raw) = raw_result {
                 raw.cert_info = rustls_result.cert_info;

src/transport/middle_proxy/config_updater.rs

@@ -228,6 +228,10 @@ async fn run_update_cycle(
         cfg.general.me_pool_drain_ttl_secs,
         cfg.general.effective_me_pool_force_close_secs(),
         cfg.general.me_pool_min_fresh_ratio,
+        cfg.general.me_hardswap_warmup_delay_min_ms,
+        cfg.general.me_hardswap_warmup_delay_max_ms,
+        cfg.general.me_hardswap_warmup_extra_passes,
+        cfg.general.me_hardswap_warmup_pass_backoff_base_ms,
     );
 
     let required_cfg_snapshots = cfg.general.me_config_stable_snapshots.max(1);
@@ -407,6 +411,10 @@ pub async fn me_config_updater(
                     cfg.general.me_pool_drain_ttl_secs,
                     cfg.general.effective_me_pool_force_close_secs(),
                     cfg.general.me_pool_min_fresh_ratio,
+                    cfg.general.me_hardswap_warmup_delay_min_ms,
+                    cfg.general.me_hardswap_warmup_delay_max_ms,
+                    cfg.general.me_hardswap_warmup_extra_passes,
+                    cfg.general.me_hardswap_warmup_pass_backoff_base_ms,
                 );
                 let new_secs = cfg.general.effective_update_every_secs().max(1);
                 if new_secs == update_every_secs {

src/transport/middle_proxy/handshake.rs

@@ -14,13 +14,16 @@ use tokio::net::{TcpStream, TcpSocket};
 use tokio::time::timeout;
 use tracing::{debug, info, warn};
 
+use crate::config::MeSocksKdfPolicy;
 use crate::crypto::{SecureRandom, build_middleproxy_prekey, derive_middleproxy_keys, sha256};
 use crate::error::{ProxyError, Result};
 use crate::network::IpFamily;
+use crate::network::probe::is_bogon;
 use crate::protocol::constants::{
     ME_CONNECT_TIMEOUT_SECS, ME_HANDSHAKE_TIMEOUT_SECS, RPC_CRYPTO_AES_U32,
     RPC_HANDSHAKE_ERROR_U32, rpc_crypto_flags,
 };
+use crate::transport::{UpstreamEgressInfo, UpstreamRouteKind};
 
 use super::codec::{
     RpcChecksumMode, build_handshake_payload, build_nonce_payload, build_rpc_frame,
@@ -43,33 +46,125 @@ pub(crate) struct HandshakeOutput {
 }
 
 impl MePool {
-    /// TCP connect with timeout + return RTT in milliseconds.
-    pub(crate) async fn connect_tcp(&self, addr: SocketAddr) -> Result<(TcpStream, f64)> {
-        let start = Instant::now();
-        let connect_fut = async {
-            if addr.is_ipv6()
-                && let Some(v6) = self.detected_ipv6
-            {
-                match TcpSocket::new_v6() {
-                    Ok(sock) => {
-                        if let Err(e) = sock.bind(SocketAddr::new(IpAddr::V6(v6), 0)) {
-                            debug!(error = %e, bind_ip = %v6, "ME IPv6 bind failed, falling back to default bind");
-                        } else {
-                            match sock.connect(addr).await {
-                                Ok(stream) => return Ok(stream),
-                                Err(e) => debug!(error = %e, target = %addr, "ME IPv6 bound connect failed, retrying default connect"),
-                            }
-                        }
+    async fn resolve_dc_idx_for_endpoint(&self, addr: SocketAddr) -> Option<i16> {
+        if addr.is_ipv4() {
+            let map = self.proxy_map_v4.read().await;
+            for (dc, addrs) in map.iter() {
+                if addrs
+                    .iter()
+                    .any(|(ip, port)| SocketAddr::new(*ip, *port) == addr)
+                {
+                    let abs_dc = dc.abs();
+                    if abs_dc > 0
+                        && let Ok(dc_idx) = i16::try_from(abs_dc)
+                    {
+                        return Some(dc_idx);
                     }
-                    Err(e) => debug!(error = %e, "ME IPv6 socket creation failed, falling back to default connect"),
                 }
             }
-            TcpStream::connect(addr).await
+        } else {
+            let map = self.proxy_map_v6.read().await;
+            for (dc, addrs) in map.iter() {
+                if addrs
+                    .iter()
+                    .any(|(ip, port)| SocketAddr::new(*ip, *port) == addr)
+                {
+                    let abs_dc = dc.abs();
+                    if abs_dc > 0
+                        && let Ok(dc_idx) = i16::try_from(abs_dc)
+                    {
+                        return Some(dc_idx);
+                    }
+                }
+            }
+        }
+        None
+    }
+
+    fn direct_bind_ip_for_stun(
+        family: IpFamily,
+        upstream_egress: Option<UpstreamEgressInfo>,
+    ) -> Option<IpAddr> {
+        let info = upstream_egress?;
+        if info.route_kind != UpstreamRouteKind::Direct {
+            return None;
+        }
+        match (family, info.direct_bind_ip) {
+            (IpFamily::V4, Some(IpAddr::V4(ip))) => Some(IpAddr::V4(ip)),
+            (IpFamily::V6, Some(IpAddr::V6(ip))) => Some(IpAddr::V6(ip)),
+            _ => None,
+        }
+    }
+
+    fn select_socks_bound_addr(
+        family: IpFamily,
+        upstream_egress: Option<UpstreamEgressInfo>,
+    ) -> Option<SocketAddr> {
+        let info = upstream_egress?;
+        if !matches!(
+            info.route_kind,
+            UpstreamRouteKind::Socks4 | UpstreamRouteKind::Socks5
+        ) {
+            return None;
+        }
+        let bound = info.socks_bound_addr?;
+        let family_matches = matches!(
+            (family, bound.ip()),
+            (IpFamily::V4, IpAddr::V4(_)) | (IpFamily::V6, IpAddr::V6(_))
+        );
+        if !family_matches || is_bogon(bound.ip()) || bound.ip().is_unspecified() {
+            return None;
+        }
+        Some(bound)
+    }
+
+    fn is_socks_route(upstream_egress: Option<UpstreamEgressInfo>) -> bool {
+        matches!(
+            upstream_egress.map(|info| info.route_kind),
+            Some(UpstreamRouteKind::Socks4 | UpstreamRouteKind::Socks5)
+        )
+    }
+
+    /// TCP connect with timeout + return RTT in milliseconds.
+    pub(crate) async fn connect_tcp(
+        &self,
+        addr: SocketAddr,
+    ) -> Result<(TcpStream, f64, Option<UpstreamEgressInfo>)> {
+        let start = Instant::now();
+        let (stream, upstream_egress) = if let Some(upstream) = &self.upstream {
+            let dc_idx = self.resolve_dc_idx_for_endpoint(addr).await;
+            let (stream, egress) = upstream.connect_with_details(addr, dc_idx, None).await?;
+            (stream, Some(egress))
+        } else {
+            let connect_fut = async {
+                if addr.is_ipv6()
+                    && let Some(v6) = self.detected_ipv6
+                {
+                    match TcpSocket::new_v6() {
+                        Ok(sock) => {
+                            if let Err(e) = sock.bind(SocketAddr::new(IpAddr::V6(v6), 0)) {
+                                debug!(error = %e, bind_ip = %v6, "ME IPv6 bind failed, falling back to default bind");
+                            } else {
+                                match sock.connect(addr).await {
+                                    Ok(stream) => return Ok(stream),
+                                    Err(e) => debug!(error = %e, target = %addr, "ME IPv6 bound connect failed, retrying default connect"),
+                                }
+                            }
+                        }
+                        Err(e) => debug!(error = %e, "ME IPv6 socket creation failed, falling back to default connect"),
+                    }
+                }
+                TcpStream::connect(addr).await
+            };
+
+            let stream = timeout(Duration::from_secs(ME_CONNECT_TIMEOUT_SECS), connect_fut)
+                .await
+                .map_err(|_| ProxyError::ConnectionTimeout {
+                    addr: addr.to_string(),
+                })??;
+            (stream, None)
         };
 
-        let stream = timeout(Duration::from_secs(ME_CONNECT_TIMEOUT_SECS), connect_fut)
-            .await
-            .map_err(|_| ProxyError::ConnectionTimeout { addr: addr.to_string() })??;
         let connect_ms = start.elapsed().as_secs_f64() * 1000.0;
         stream.set_nodelay(true).ok();
         if let Err(e) = Self::configure_keepalive(&stream) {
@@ -79,7 +174,7 @@ impl MePool {
         if let Err(e) = Self::configure_user_timeout(stream.as_raw_fd()) {
             warn!(error = %e, "ME TCP_USER_TIMEOUT setup failed");
         }
-        Ok((stream, connect_ms))
+        Ok((stream, connect_ms, upstream_egress))
     }
 
     fn configure_keepalive(stream: &TcpStream) -> std::io::Result<()> {
@@ -117,12 +212,14 @@ impl MePool {
         &self,
         stream: TcpStream,
         addr: SocketAddr,
+        upstream_egress: Option<UpstreamEgressInfo>,
         rng: &SecureRandom,
     ) -> Result<HandshakeOutput> {
         let hs_start = Instant::now();
 
         let local_addr = stream.local_addr().map_err(ProxyError::Io)?;
-        let peer_addr = stream.peer_addr().map_err(ProxyError::Io)?;
+        let transport_peer_addr = stream.peer_addr().map_err(ProxyError::Io)?;
+        let peer_addr = addr;
 
         let _ = self.maybe_detect_nat_ip(local_addr.ip()).await;
         let family = if local_addr.ip().is_ipv4() {
@@ -130,8 +227,32 @@ impl MePool {
         } else {
             IpFamily::V6
         };
-        let reflected = if self.nat_probe {
-            self.maybe_reflect_public_addr(family).await
+        let is_socks_route = Self::is_socks_route(upstream_egress);
+        let socks_bound_addr = Self::select_socks_bound_addr(family, upstream_egress);
+        let reflected = if let Some(bound) = socks_bound_addr {
+            Some(bound)
+        } else if is_socks_route {
+            match self.socks_kdf_policy() {
+                MeSocksKdfPolicy::Strict => {
+                    self.stats.increment_me_socks_kdf_strict_reject();
+                    return Err(ProxyError::InvalidHandshake(
+                        "SOCKS route returned no valid BND.ADDR for ME KDF (strict policy)"
+                            .to_string(),
+                    ));
+                }
+                MeSocksKdfPolicy::Compat => {
+                    self.stats.increment_me_socks_kdf_compat_fallback();
+                    if self.nat_probe {
+                        let bind_ip = Self::direct_bind_ip_for_stun(family, upstream_egress);
+                        self.maybe_reflect_public_addr(family, bind_ip).await
+                    } else {
+                        None
+                    }
+                }
+            }
+        } else if self.nat_probe {
+            let bind_ip = Self::direct_bind_ip_for_stun(family, upstream_egress);
+            self.maybe_reflect_public_addr(family, bind_ip).await
         } else {
             None
         };
@@ -197,7 +318,9 @@ impl MePool {
             %local_addr_nat,
             reflected_ip = reflected.map(|r| r.ip()).as_ref().map(ToString::to_string),
             %peer_addr,
+            %transport_peer_addr,
             %peer_addr_nat,
+            socks_bound_addr = socks_bound_addr.map(|v| v.to_string()),
             key_selector = format_args!("0x{ks:08x}"),
             crypto_schema = format_args!("0x{schema:08x}"),
             skew_secs = skew,
@@ -206,7 +329,11 @@ impl MePool {
 
         let ts_bytes = crypto_ts.to_le_bytes();
         let server_port_bytes = peer_addr_nat.port().to_le_bytes();
-        let client_port_bytes = local_addr_nat.port().to_le_bytes();
+        let client_port_for_kdf = socks_bound_addr
+            .map(|bound| bound.port())
+            .filter(|port| *port != 0)
+            .unwrap_or(local_addr_nat.port());
+        let client_port_bytes = client_port_for_kdf.to_le_bytes();
 
         let server_ip = extract_ip_material(peer_addr_nat);
         let client_ip = extract_ip_material(local_addr_nat);

src/transport/middle_proxy/health.rs

@@ -1,10 +1,9 @@
-use std::collections::{HashMap, HashSet};
+use std::collections::HashMap;
 use std::net::SocketAddr;
 use std::sync::Arc;
 use std::time::{Duration, Instant};
 
 use tracing::{debug, info, warn};
-use rand::seq::SliceRandom;
 use rand::Rng;
 
 use crate::crypto::SecureRandom;
@@ -23,6 +22,7 @@ pub async fn me_health_monitor(pool: Arc<MePool>, rng: Arc<SecureRandom>, _min_c
     let mut inflight: HashMap<(i32, IpFamily), usize> = HashMap::new();
     loop {
         tokio::time::sleep(Duration::from_secs(HEALTH_INTERVAL_SECS)).await;
+        pool.prune_closed_writers().await;
         check_family(
             IpFamily::V4,
             &pool,
@@ -64,31 +64,43 @@ async fn check_family(
         IpFamily::V4 => pool.proxy_map_v4.read().await.clone(),
         IpFamily::V6 => pool.proxy_map_v6.read().await.clone(),
     };
-    let writer_addrs: HashSet<SocketAddr> = pool
+
+    let mut dc_endpoints = HashMap::<i32, Vec<SocketAddr>>::new();
+    for (dc, addrs) in map {
+        let entry = dc_endpoints.entry(dc.abs()).or_default();
+        for (ip, port) in addrs {
+            entry.push(SocketAddr::new(ip, port));
+        }
+    }
+    for endpoints in dc_endpoints.values_mut() {
+        endpoints.sort_unstable();
+        endpoints.dedup();
+    }
+
+    let mut live_addr_counts = HashMap::<SocketAddr, usize>::new();
+    for writer in pool
         .writers
         .read()
         .await
         .iter()
         .filter(|w| !w.draining.load(std::sync::atomic::Ordering::Relaxed))
-        .map(|w| w.addr)
-        .collect();
+    {
+        *live_addr_counts.entry(writer.addr).or_insert(0) += 1;
+    }
 
-    let entries: Vec<(i32, Vec<SocketAddr>)> = map
-        .iter()
-        .map(|(dc, addrs)| {
-            let list = addrs
-                .iter()
-                .map(|(ip, port)| SocketAddr::new(*ip, *port))
-                .collect::<Vec<_>>();
-            (*dc, list)
-        })
-        .collect();
-
-    for (dc, dc_addrs) in entries {
-        let has_coverage = dc_addrs.iter().any(|a| writer_addrs.contains(a));
-        if has_coverage {
+    for (dc, endpoints) in dc_endpoints {
+        if endpoints.is_empty() {
             continue;
         }
+        let required = MePool::required_writers_for_dc(endpoints.len());
+        let alive = endpoints
+            .iter()
+            .map(|addr| *live_addr_counts.get(addr).unwrap_or(&0))
+            .sum::<usize>();
+        if alive >= required {
+            continue;
+        }
+        let missing = required - alive;
 
         let key = (dc, family);
         let now = Instant::now();
@@ -104,32 +116,45 @@ async fn check_family(
         }
         *inflight.entry(key).or_insert(0) += 1;
 
-        let mut shuffled = dc_addrs.clone();
-        shuffled.shuffle(&mut rand::rng());
-        let mut success = false;
-        for addr in shuffled {
-            let res = tokio::time::timeout(pool.me_one_timeout, pool.connect_one(addr, rng.as_ref())).await;
+        let mut restored = 0usize;
+        for _ in 0..missing {
+            let res = tokio::time::timeout(
+                pool.me_one_timeout,
+                pool.connect_endpoints_round_robin(&endpoints, rng.as_ref()),
+            )
+            .await;
             match res {
-                Ok(Ok(())) => {
-                    info!(%addr, dc = %dc, ?family, "ME reconnected for DC coverage");
+                Ok(true) => {
+                    restored += 1;
                     pool.stats.increment_me_reconnect_success();
-                    backoff.insert(key, pool.me_reconnect_backoff_base.as_millis() as u64);
-                    let jitter = pool.me_reconnect_backoff_base.as_millis() as u64 / JITTER_FRAC_NUM;
-                    let wait = pool.me_reconnect_backoff_base
-                        + Duration::from_millis(rand::rng().random_range(0..=jitter.max(1)));
-                    next_attempt.insert(key, now + wait);
-                    success = true;
-                    break;
                 }
-                Ok(Err(e)) => {
+                Ok(false) => {
                     pool.stats.increment_me_reconnect_attempt();
-                    debug!(%addr, dc = %dc, error = %e, ?family, "ME reconnect failed")
+                    debug!(dc = %dc, ?family, "ME round-robin reconnect failed")
+                }
+                Err(_) => {
+                    pool.stats.increment_me_reconnect_attempt();
+                    debug!(dc = %dc, ?family, "ME reconnect timed out");
                 }
-                Err(_) => debug!(%addr, dc = %dc, ?family, "ME reconnect timed out"),
             }
         }
-        if !success {
-            pool.stats.increment_me_reconnect_attempt();
+
+        let now_alive = alive + restored;
+        if now_alive >= required {
+            info!(
+                dc = %dc,
+                ?family,
+                alive = now_alive,
+                required,
+                endpoint_count = endpoints.len(),
+                "ME writer floor restored for DC"
+            );
+            backoff.insert(key, pool.me_reconnect_backoff_base.as_millis() as u64);
+            let jitter = pool.me_reconnect_backoff_base.as_millis() as u64 / JITTER_FRAC_NUM;
+            let wait = pool.me_reconnect_backoff_base
+                + Duration::from_millis(rand::rng().random_range(0..=jitter.max(1)));
+            next_attempt.insert(key, now + wait);
+        } else {
             let curr = *backoff.get(&key).unwrap_or(&(pool.me_reconnect_backoff_base.as_millis() as u64));
             let next_ms = (curr.saturating_mul(2)).min(pool.me_reconnect_backoff_cap.as_millis() as u64);
             backoff.insert(key, next_ms);
@@ -137,7 +162,15 @@ async fn check_family(
             let wait = Duration::from_millis(next_ms)
                 + Duration::from_millis(rand::rng().random_range(0..=jitter.max(1)));
             next_attempt.insert(key, now + wait);
-            warn!(dc = %dc, backoff_ms = next_ms, ?family, "DC has no ME coverage, scheduled reconnect");
+            warn!(
+                dc = %dc,
+                ?family,
+                alive = now_alive,
+                required,
+                endpoint_count = endpoints.len(),
+                backoff_ms = next_ms,
+                "DC writer floor is below required level, scheduled reconnect"
+            );
         }
         if let Some(v) = inflight.get_mut(&key) {
             *v = v.saturating_sub(1);

src/transport/middle_proxy/mod.rs

@@ -1,24 +1,29 @@
 //! Middle Proxy RPC transport.
 
 mod codec;
+mod config_updater;
 mod handshake;
 mod health;
 mod pool;
+mod pool_config;
+mod pool_init;
 mod pool_nat;
+mod pool_refill;
+mod pool_reinit;
+mod pool_writer;
 mod ping;
 mod reader;
 mod registry;
+mod rotation;
 mod send;
 mod secret;
-mod rotation;
-mod config_updater;
 mod wire;
 
 use bytes::Bytes;
 
 pub use health::me_health_monitor;
 #[allow(unused_imports)]
-pub use ping::{run_me_ping, format_sample_line, MePingReport, MePingSample, MePingFamily};
+pub use ping::{run_me_ping, format_sample_line, format_me_route, MePingReport, MePingSample, MePingFamily};
 pub use pool::MePool;
 #[allow(unused_imports)]
 pub use pool_nat::{stun_probe, detect_public_ip};

src/transport/middle_proxy/ping.rs

@@ -2,8 +2,12 @@ use std::collections::HashMap;
 use std::net::{IpAddr, SocketAddr};
 use std::sync::Arc;
 
+use tokio::net::UdpSocket;
+
+use crate::config::{UpstreamConfig, UpstreamType};
 use crate::crypto::SecureRandom;
 use crate::error::ProxyError;
+use crate::transport::{UpstreamEgressInfo, UpstreamRouteKind};
 
 use super::MePool;
 
@@ -17,6 +21,7 @@ pub enum MePingFamily {
 pub struct MePingSample {
     pub dc: i32,
     pub addr: SocketAddr,
+    pub route: Option<String>,
     pub connect_ms: Option<f64>,
     pub handshake_ms: Option<f64>,
     pub error: Option<String>,
@@ -50,6 +55,208 @@ pub fn format_sample_line(sample: &MePingSample) -> String {
     }
 }
 
+fn format_direct_with_config(
+    interface: &Option<String>,
+    bind_addresses: &Option<Vec<String>>,
+) -> Option<String> {
+    let mut direct_parts: Vec<String> = Vec::new();
+    if let Some(dev) = interface.as_deref().filter(|v| !v.is_empty()) {
+        direct_parts.push(format!("dev={dev}"));
+    }
+    if let Some(src) = bind_addresses.as_ref().filter(|v| !v.is_empty()) {
+        direct_parts.push(format!("src={}", src.join(",")));
+    }
+    if direct_parts.is_empty() {
+        None
+    } else {
+        Some(format!("direct {}", direct_parts.join(" ")))
+    }
+}
+
+fn pick_target_for_family(reports: &[MePingReport], family: MePingFamily) -> Option<SocketAddr> {
+    reports.iter().find_map(|report| {
+        if report.family != family {
+            return None;
+        }
+        report
+            .samples
+            .iter()
+            .find(|s| s.error.is_none() && s.handshake_ms.is_some())
+            .map(|s| s.addr)
+    })
+}
+
+fn route_from_egress(egress: Option<UpstreamEgressInfo>) -> Option<String> {
+    let info = egress?;
+    match info.route_kind {
+        UpstreamRouteKind::Direct => {
+            let src_ip = info
+                .direct_bind_ip
+                .or_else(|| info.local_addr.map(|addr| addr.ip()));
+            let ip = src_ip?;
+            let mut parts = Vec::new();
+            if let Some(dev) = detect_interface_for_ip(ip) {
+                parts.push(format!("dev={dev}"));
+            }
+            parts.push(format!("src={ip}"));
+            Some(format!("direct {}", parts.join(" ")))
+        }
+        UpstreamRouteKind::Socks4 => {
+            let route = info
+                .socks_proxy_addr
+                .map(|addr| format!("socks4://{addr}"))
+                .unwrap_or_else(|| "socks4://unknown".to_string());
+            Some(match info.socks_bound_addr {
+                Some(bound) => format!("{route} bnd={bound}"),
+                None => route,
+            })
+        }
+        UpstreamRouteKind::Socks5 => {
+            let route = info
+                .socks_proxy_addr
+                .map(|addr| format!("socks5://{addr}"))
+                .unwrap_or_else(|| "socks5://unknown".to_string());
+            Some(match info.socks_bound_addr {
+                Some(bound) => format!("{route} bnd={bound}"),
+                None => route,
+            })
+        }
+    }
+}
+
+#[cfg(unix)]
+fn detect_interface_for_ip(ip: IpAddr) -> Option<String> {
+    use nix::ifaddrs::getifaddrs;
+
+    if let Ok(addrs) = getifaddrs() {
+        for iface in addrs {
+            if let Some(address) = iface.address {
+                if let Some(v4) = address.as_sockaddr_in() {
+                    if IpAddr::V4(v4.ip()) == ip {
+                        return Some(iface.interface_name);
+                    }
+                } else if let Some(v6) = address.as_sockaddr_in6() {
+                    if IpAddr::V6(v6.ip()) == ip {
+                        return Some(iface.interface_name);
+                    }
+                }
+            }
+        }
+    }
+    None
+}
+
+#[cfg(not(unix))]
+fn detect_interface_for_ip(_ip: IpAddr) -> Option<String> {
+    None
+}
+
+async fn detect_direct_route_details(
+    reports: &[MePingReport],
+    prefer_ipv6: bool,
+    v4_ok: bool,
+    v6_ok: bool,
+) -> Option<String> {
+    let target_addr = if prefer_ipv6 && v6_ok {
+        pick_target_for_family(reports, MePingFamily::V6)
+            .or_else(|| pick_target_for_family(reports, MePingFamily::V4))
+    } else if v4_ok {
+        pick_target_for_family(reports, MePingFamily::V4)
+            .or_else(|| pick_target_for_family(reports, MePingFamily::V6))
+    } else {
+        pick_target_for_family(reports, MePingFamily::V6)
+            .or_else(|| pick_target_for_family(reports, MePingFamily::V4))
+    }?;
+
+    let local_ip = if target_addr.is_ipv4() {
+        let sock = UdpSocket::bind("0.0.0.0:0").await.ok()?;
+        sock.connect(target_addr).await.ok()?;
+        sock.local_addr().ok().map(|a| a.ip())
+    } else {
+        let sock = UdpSocket::bind("[::]:0").await.ok()?;
+        sock.connect(target_addr).await.ok()?;
+        sock.local_addr().ok().map(|a| a.ip())
+    };
+
+    let mut parts = Vec::new();
+    if let Some(ip) = local_ip {
+        if let Some(dev) = detect_interface_for_ip(ip) {
+            parts.push(format!("dev={dev}"));
+        }
+        parts.push(format!("src={ip}"));
+    }
+
+    if parts.is_empty() {
+        None
+    } else {
+        Some(format!("direct {}", parts.join(" ")))
+    }
+}
+
+pub async fn format_me_route(
+    upstreams: &[UpstreamConfig],
+    reports: &[MePingReport],
+    prefer_ipv6: bool,
+    v4_ok: bool,
+    v6_ok: bool,
+) -> String {
+    if let Some(route) = reports
+        .iter()
+        .flat_map(|report| report.samples.iter())
+        .find(|sample| sample.error.is_none() && sample.handshake_ms.is_some())
+        .and_then(|sample| sample.route.clone())
+    {
+        return route;
+    }
+
+    let enabled_upstreams: Vec<_> = upstreams.iter().filter(|u| u.enabled).collect();
+    if enabled_upstreams.is_empty() {
+        return detect_direct_route_details(reports, prefer_ipv6, v4_ok, v6_ok)
+            .await
+            .unwrap_or_else(|| "direct".to_string());
+    }
+
+    if enabled_upstreams.len() == 1 {
+        return match &enabled_upstreams[0].upstream_type {
+            UpstreamType::Direct {
+                interface,
+                bind_addresses,
+            } => {
+                if let Some(route) = format_direct_with_config(interface, bind_addresses) {
+                    route
+                } else {
+                    detect_direct_route_details(reports, prefer_ipv6, v4_ok, v6_ok)
+                        .await
+                        .unwrap_or_else(|| "direct".to_string())
+                }
+            }
+            UpstreamType::Socks4 { address, .. } => format!("socks4://{address}"),
+            UpstreamType::Socks5 { address, .. } => format!("socks5://{address}"),
+        };
+    }
+
+    let has_direct = enabled_upstreams
+        .iter()
+        .any(|u| matches!(u.upstream_type, UpstreamType::Direct { .. }));
+    let has_socks4 = enabled_upstreams
+        .iter()
+        .any(|u| matches!(u.upstream_type, UpstreamType::Socks4 { .. }));
+    let has_socks5 = enabled_upstreams
+        .iter()
+        .any(|u| matches!(u.upstream_type, UpstreamType::Socks5 { .. }));
+    let mut kinds = Vec::new();
+    if has_direct {
+        kinds.push("direct");
+    }
+    if has_socks4 {
+        kinds.push("socks4");
+    }
+    if has_socks5 {
+        kinds.push("socks5");
+    }
+    format!("mixed upstreams ({})", kinds.join(", "))
+}
+
 #[cfg(test)]
 mod tests {
     use super::*;
@@ -64,6 +271,7 @@ mod tests {
         let s = sample(MePingSample {
             dc: 4,
             addr: SocketAddr::new(IpAddr::V4(Ipv4Addr::new(1, 2, 3, 4)), 8888),
+            route: Some("direct src=1.2.3.4".to_string()),
             connect_ms: Some(12.3),
             handshake_ms: Some(34.7),
             error: None,
@@ -80,6 +288,7 @@ mod tests {
         let s = sample(MePingSample {
             dc: -5,
             addr: SocketAddr::new(IpAddr::V4(Ipv4Addr::new(5, 6, 7, 8)), 80),
+            route: Some("socks5".to_string()),
             connect_ms: Some(10.0),
             handshake_ms: None,
             error: Some("handshake timeout".to_string()),
@@ -120,11 +329,13 @@ pub async fn run_me_ping(pool: &Arc<MePool>, rng: &SecureRandom) -> Vec<MePingRe
             let mut connect_ms = None;
             let mut handshake_ms = None;
             let mut error = None;
+            let mut route = None;
 
             match pool.connect_tcp(addr).await {
-                Ok((stream, conn_rtt)) => {
+                Ok((stream, conn_rtt, upstream_egress)) => {
                     connect_ms = Some(conn_rtt);
-                    match pool.handshake_only(stream, addr, rng).await {
+                    route = route_from_egress(upstream_egress);
+                    match pool.handshake_only(stream, addr, upstream_egress, rng).await {
                         Ok(hs) => {
                             handshake_ms = Some(hs.handshake_ms);
                             // drop halves to close
@@ -144,6 +355,7 @@ pub async fn run_me_ping(pool: &Arc<MePool>, rng: &SecureRandom) -> Vec<MePingRe
             samples.push(MePingSample {
                 dc,
                 addr,
+                route,
                 connect_ms,
                 handshake_ms,
                 error,

src/transport/middle_proxy/pool.rs

@@ -1,27 +1,20 @@
 use std::collections::{HashMap, HashSet};
 use std::net::{IpAddr, Ipv6Addr, SocketAddr};
 use std::sync::Arc;
-use std::sync::atomic::{AtomicBool, AtomicI32, AtomicU32, AtomicU64, AtomicUsize, Ordering};
-use bytes::BytesMut;
-use rand::Rng;
-use rand::seq::SliceRandom;
-use tokio::sync::{Mutex, RwLock, mpsc, Notify};
-use tokio_util::sync::CancellationToken;
-use tracing::{debug, info, warn};
+use std::sync::atomic::{AtomicBool, AtomicI32, AtomicU8, AtomicU32, AtomicU64, AtomicUsize, Ordering};
 use std::time::{Duration, Instant, SystemTime, UNIX_EPOCH};
 
+use tokio::sync::{Mutex, Notify, RwLock, mpsc};
+use tokio_util::sync::CancellationToken;
+
+use crate::config::MeSocksKdfPolicy;
 use crate::crypto::SecureRandom;
-use crate::error::{ProxyError, Result};
-use crate::network::probe::NetworkDecision;
 use crate::network::IpFamily;
-use crate::protocol::constants::*;
+use crate::network::probe::NetworkDecision;
+use crate::transport::UpstreamManager;
 
 use super::ConnRegistry;
-use super::registry::BoundConn;
-use super::codec::{RpcWriter, WriterCommand};
-use super::reader::reader_loop;
-const ME_ACTIVE_PING_SECS: u64 = 25;
-const ME_ACTIVE_PING_JITTER_SECS: i64 = 5;
+use super::codec::WriterCommand;
 
 #[derive(Clone)]
 pub struct MeWriter {
@@ -42,6 +35,7 @@ pub struct MePool {
     pub(super) writers: Arc<RwLock<Vec<MeWriter>>>,
     pub(super) rr: AtomicU64,
     pub(super) decision: NetworkDecision,
+    pub(super) upstream: Option<Arc<UpstreamManager>>,
     pub(super) rng: Arc<SecureRandom>,
     pub(super) proxy_tag: Option<Vec<u8>>,
     pub(super) proxy_secret: Arc<RwLock<Vec<u8>>>,
@@ -50,6 +44,8 @@ pub struct MePool {
     pub(super) nat_probe: bool,
     pub(super) nat_stun: Option<String>,
     pub(super) nat_stun_servers: Vec<String>,
+    pub(super) nat_stun_live_servers: Arc<RwLock<Vec<String>>>,
+    pub(super) nat_probe_concurrency: usize,
     pub(super) detected_ipv6: Option<Ipv6Addr>,
     pub(super) nat_probe_attempts: std::sync::atomic::AtomicU8,
     pub(super) nat_probe_disabled: std::sync::atomic::AtomicBool,
@@ -75,6 +71,7 @@ pub struct MePool {
     pub(super) rtt_stats: Arc<Mutex<HashMap<u64, (f64, f64)>>>,
     pub(super) nat_reflection_cache: Arc<Mutex<NatReflectionCache>>,
     pub(super) writer_available: Arc<Notify>,
+    pub(super) refill_inflight: Arc<Mutex<HashSet<SocketAddr>>>,
     pub(super) conn_count: AtomicUsize,
     pub(super) stats: Arc<crate::stats::Stats>,
     pub(super) generation: AtomicU64,
@@ -82,6 +79,11 @@ pub struct MePool {
     pub(super) me_pool_drain_ttl_secs: AtomicU64,
     pub(super) me_pool_force_close_secs: AtomicU64,
     pub(super) me_pool_min_fresh_ratio_permille: AtomicU32,
+    pub(super) me_hardswap_warmup_delay_min_ms: AtomicU64,
+    pub(super) me_hardswap_warmup_delay_max_ms: AtomicU64,
+    pub(super) me_hardswap_warmup_extra_passes: AtomicU32,
+    pub(super) me_hardswap_warmup_pass_backoff_base_ms: AtomicU64,
+    pub(super) me_socks_kdf_policy: AtomicU8,
     pool_size: usize,
 }
 
@@ -97,11 +99,11 @@ impl MePool {
         (clamped * 1000.0).round() as u32
     }
 
-    fn permille_to_ratio(permille: u32) -> f32 {
+    pub(super) fn permille_to_ratio(permille: u32) -> f32 {
         (permille.min(1000) as f32) / 1000.0
     }
 
-    fn now_epoch_secs() -> u64 {
+    pub(super) fn now_epoch_secs() -> u64 {
         SystemTime::now()
             .duration_since(UNIX_EPOCH)
             .unwrap_or_default()
@@ -115,6 +117,7 @@ impl MePool {
         nat_probe: bool,
         nat_stun: Option<String>,
         nat_stun_servers: Vec<String>,
+        nat_probe_concurrency: usize,
         detected_ipv6: Option<Ipv6Addr>,
         me_one_retry: u8,
         me_one_timeout_ms: u64,
@@ -122,6 +125,7 @@ impl MePool {
         proxy_map_v6: HashMap<i32, Vec<(IpAddr, u16)>>,
         default_dc: Option<i32>,
         decision: NetworkDecision,
+        upstream: Option<Arc<UpstreamManager>>,
         rng: Arc<SecureRandom>,
         stats: Arc<crate::stats::Stats>,
         me_keepalive_enabled: bool,
@@ -139,12 +143,27 @@ impl MePool {
         me_pool_drain_ttl_secs: u64,
         me_pool_force_close_secs: u64,
         me_pool_min_fresh_ratio: f32,
+        me_hardswap_warmup_delay_min_ms: u64,
+        me_hardswap_warmup_delay_max_ms: u64,
+        me_hardswap_warmup_extra_passes: u8,
+        me_hardswap_warmup_pass_backoff_base_ms: u64,
+        me_socks_kdf_policy: MeSocksKdfPolicy,
+        me_route_backpressure_base_timeout_ms: u64,
+        me_route_backpressure_high_timeout_ms: u64,
+        me_route_backpressure_high_watermark_pct: u8,
     ) -> Arc<Self> {
+        let registry = Arc::new(ConnRegistry::new());
+        registry.update_route_backpressure_policy(
+            me_route_backpressure_base_timeout_ms,
+            me_route_backpressure_high_timeout_ms,
+            me_route_backpressure_high_watermark_pct,
+        );
         Arc::new(Self {
-            registry: Arc::new(ConnRegistry::new()),
+            registry,
             writers: Arc::new(RwLock::new(Vec::new())),
             rr: AtomicU64::new(0),
             decision,
+            upstream,
             rng,
             proxy_tag,
             proxy_secret: Arc::new(RwLock::new(proxy_secret)),
@@ -153,6 +172,8 @@ impl MePool {
             nat_probe,
             nat_stun,
             nat_stun_servers,
+            nat_stun_live_servers: Arc::new(RwLock::new(Vec::new())),
+            nat_probe_concurrency: nat_probe_concurrency.max(1),
             detected_ipv6,
             nat_probe_attempts: std::sync::atomic::AtomicU8::new(0),
             nat_probe_disabled: std::sync::atomic::AtomicBool::new(false),
@@ -180,12 +201,22 @@ impl MePool {
             rtt_stats: Arc::new(Mutex::new(HashMap::new())),
             nat_reflection_cache: Arc::new(Mutex::new(NatReflectionCache::default())),
             writer_available: Arc::new(Notify::new()),
+            refill_inflight: Arc::new(Mutex::new(HashSet::new())),
             conn_count: AtomicUsize::new(0),
             generation: AtomicU64::new(1),
             hardswap: AtomicBool::new(hardswap),
             me_pool_drain_ttl_secs: AtomicU64::new(me_pool_drain_ttl_secs),
             me_pool_force_close_secs: AtomicU64::new(me_pool_force_close_secs),
-            me_pool_min_fresh_ratio_permille: AtomicU32::new(Self::ratio_to_permille(me_pool_min_fresh_ratio)),
+            me_pool_min_fresh_ratio_permille: AtomicU32::new(Self::ratio_to_permille(
+                me_pool_min_fresh_ratio,
+            )),
+            me_hardswap_warmup_delay_min_ms: AtomicU64::new(me_hardswap_warmup_delay_min_ms),
+            me_hardswap_warmup_delay_max_ms: AtomicU64::new(me_hardswap_warmup_delay_max_ms),
+            me_hardswap_warmup_extra_passes: AtomicU32::new(me_hardswap_warmup_extra_passes as u32),
+            me_hardswap_warmup_pass_backoff_base_ms: AtomicU64::new(
+                me_hardswap_warmup_pass_backoff_base_ms,
+            ),
+            me_socks_kdf_policy: AtomicU8::new(me_socks_kdf_policy.as_u8()),
         })
     }
 
@@ -203,18 +234,34 @@ impl MePool {
         drain_ttl_secs: u64,
         force_close_secs: u64,
         min_fresh_ratio: f32,
+        hardswap_warmup_delay_min_ms: u64,
+        hardswap_warmup_delay_max_ms: u64,
+        hardswap_warmup_extra_passes: u8,
+        hardswap_warmup_pass_backoff_base_ms: u64,
     ) {
         self.hardswap.store(hardswap, Ordering::Relaxed);
-        self.me_pool_drain_ttl_secs.store(drain_ttl_secs, Ordering::Relaxed);
+        self.me_pool_drain_ttl_secs
+            .store(drain_ttl_secs, Ordering::Relaxed);
         self.me_pool_force_close_secs
             .store(force_close_secs, Ordering::Relaxed);
         self.me_pool_min_fresh_ratio_permille
             .store(Self::ratio_to_permille(min_fresh_ratio), Ordering::Relaxed);
+        self.me_hardswap_warmup_delay_min_ms
+            .store(hardswap_warmup_delay_min_ms, Ordering::Relaxed);
+        self.me_hardswap_warmup_delay_max_ms
+            .store(hardswap_warmup_delay_max_ms, Ordering::Relaxed);
+        self.me_hardswap_warmup_extra_passes
+            .store(hardswap_warmup_extra_passes as u32, Ordering::Relaxed);
+        self.me_hardswap_warmup_pass_backoff_base_ms
+            .store(hardswap_warmup_pass_backoff_base_ms, Ordering::Relaxed);
     }
 
     pub fn reset_stun_state(&self) {
         self.nat_probe_attempts.store(0, Ordering::Relaxed);
         self.nat_probe_disabled.store(false, Ordering::Relaxed);
+        if let Ok(mut live) = self.nat_stun_live_servers.try_write() {
+            live.clear();
+        }
     }
 
     pub fn translate_our_addr(&self, addr: SocketAddr) -> SocketAddr {
@@ -226,11 +273,31 @@ impl MePool {
         &self.registry
     }
 
-    fn writers_arc(&self) -> Arc<RwLock<Vec<MeWriter>>> {
+    pub fn update_runtime_transport_policy(
+        &self,
+        socks_kdf_policy: MeSocksKdfPolicy,
+        route_backpressure_base_timeout_ms: u64,
+        route_backpressure_high_timeout_ms: u64,
+        route_backpressure_high_watermark_pct: u8,
+    ) {
+        self.me_socks_kdf_policy
+            .store(socks_kdf_policy.as_u8(), Ordering::Relaxed);
+        self.registry.update_route_backpressure_policy(
+            route_backpressure_base_timeout_ms,
+            route_backpressure_high_timeout_ms,
+            route_backpressure_high_watermark_pct,
+        );
+    }
+
+    pub(super) fn socks_kdf_policy(&self) -> MeSocksKdfPolicy {
+        MeSocksKdfPolicy::from_u8(self.me_socks_kdf_policy.load(Ordering::Relaxed))
+    }
+
+    pub(super) fn writers_arc(&self) -> Arc<RwLock<Vec<MeWriter>>> {
         self.writers.clone()
     }
 
-    fn force_close_timeout(&self) -> Option<Duration> {
+    pub(super) fn force_close_timeout(&self) -> Option<Duration> {
         let secs = self.me_pool_force_close_secs.load(Ordering::Relaxed);
         if secs == 0 {
             None
@@ -239,311 +306,6 @@ impl MePool {
         }
     }
 
-    fn coverage_ratio(
-        desired_by_dc: &HashMap<i32, HashSet<SocketAddr>>,
-        active_writer_addrs: &HashSet<SocketAddr>,
-    ) -> (f32, Vec<i32>) {
-        if desired_by_dc.is_empty() {
-            return (1.0, Vec::new());
-        }
-
-        let mut missing_dc = Vec::<i32>::new();
-        let mut covered = 0usize;
-        for (dc, endpoints) in desired_by_dc {
-            if endpoints.is_empty() {
-                continue;
-            }
-            if endpoints.iter().any(|addr| active_writer_addrs.contains(addr)) {
-                covered += 1;
-            } else {
-                missing_dc.push(*dc);
-            }
-        }
-
-        missing_dc.sort_unstable();
-        let total = desired_by_dc.len().max(1);
-        let ratio = (covered as f32) / (total as f32);
-        (ratio, missing_dc)
-    }
-
-    pub async fn reconcile_connections(self: &Arc<Self>, rng: &SecureRandom) {
-        let writers = self.writers.read().await;
-        let current: HashSet<SocketAddr> = writers
-            .iter()
-            .filter(|w| !w.draining.load(Ordering::Relaxed))
-            .map(|w| w.addr)
-            .collect();
-        drop(writers);
-
-        for family in self.family_order() {
-            let map = self.proxy_map_for_family(family).await;
-            for (_dc, addrs) in map.iter() {
-                let dc_addrs: Vec<SocketAddr> = addrs
-                    .iter()
-                    .map(|(ip, port)| SocketAddr::new(*ip, *port))
-                    .collect();
-                if !dc_addrs.iter().any(|a| current.contains(a)) {
-                    let mut shuffled = dc_addrs.clone();
-                    shuffled.shuffle(&mut rand::rng());
-                    for addr in shuffled {
-                        if self.connect_one(addr, rng).await.is_ok() {
-                            break;
-                        }
-                    }
-                }
-            }
-            if !self.decision.effective_multipath && !current.is_empty() {
-                break;
-            }
-        }
-    }
-
-    async fn desired_dc_endpoints(&self) -> HashMap<i32, HashSet<SocketAddr>> {
-        let mut out: HashMap<i32, HashSet<SocketAddr>> = HashMap::new();
-
-        if self.decision.ipv4_me {
-            let map_v4 = self.proxy_map_v4.read().await.clone();
-            for (dc, addrs) in map_v4 {
-                let entry = out.entry(dc.abs()).or_default();
-                for (ip, port) in addrs {
-                    entry.insert(SocketAddr::new(ip, port));
-                }
-            }
-        }
-
-        if self.decision.ipv6_me {
-            let map_v6 = self.proxy_map_v6.read().await.clone();
-            for (dc, addrs) in map_v6 {
-                let entry = out.entry(dc.abs()).or_default();
-                for (ip, port) in addrs {
-                    entry.insert(SocketAddr::new(ip, port));
-                }
-            }
-        }
-
-        out
-    }
-
-    async fn warmup_generation_for_all_dcs(
-        self: &Arc<Self>,
-        rng: &SecureRandom,
-        generation: u64,
-        desired_by_dc: &HashMap<i32, HashSet<SocketAddr>>,
-    ) {
-        for endpoints in desired_by_dc.values() {
-            if endpoints.is_empty() {
-                continue;
-            }
-
-            let has_fresh = {
-                let ws = self.writers.read().await;
-                ws.iter().any(|w| {
-                    !w.draining.load(Ordering::Relaxed)
-                        && w.generation == generation
-                        && endpoints.contains(&w.addr)
-                })
-            };
-
-            if has_fresh {
-                continue;
-            }
-
-            let mut shuffled: Vec<SocketAddr> = endpoints.iter().copied().collect();
-            shuffled.shuffle(&mut rand::rng());
-            for addr in shuffled {
-                if self.connect_one(addr, rng).await.is_ok() {
-                    break;
-                }
-            }
-        }
-    }
-
-    pub async fn zero_downtime_reinit_after_map_change(
-        self: &Arc<Self>,
-        rng: &SecureRandom,
-    ) {
-        let desired_by_dc = self.desired_dc_endpoints().await;
-        if desired_by_dc.is_empty() {
-            warn!("ME endpoint map is empty after update; skipping stale writer drain");
-            return;
-        }
-
-        let previous_generation = self.current_generation();
-        let generation = self.generation.fetch_add(1, Ordering::Relaxed) + 1;
-        let hardswap = self.hardswap.load(Ordering::Relaxed);
-
-        if hardswap {
-            self.warmup_generation_for_all_dcs(rng, generation, &desired_by_dc)
-                .await;
-        } else {
-            self.reconcile_connections(rng).await;
-        }
-
-        let writers = self.writers.read().await;
-        let active_writer_addrs: HashSet<SocketAddr> = writers
-            .iter()
-            .filter(|w| !w.draining.load(Ordering::Relaxed))
-            .map(|w| w.addr)
-            .collect();
-        let min_ratio = Self::permille_to_ratio(
-            self.me_pool_min_fresh_ratio_permille
-                .load(Ordering::Relaxed),
-        );
-        let (coverage_ratio, missing_dc) = Self::coverage_ratio(&desired_by_dc, &active_writer_addrs);
-        if !hardswap && coverage_ratio < min_ratio {
-            warn!(
-                previous_generation,
-                generation,
-                coverage_ratio = format_args!("{coverage_ratio:.3}"),
-                min_ratio = format_args!("{min_ratio:.3}"),
-                missing_dc = ?missing_dc,
-                "ME reinit coverage below threshold; keeping stale writers"
-            );
-            return;
-        }
-
-        if hardswap {
-            let fresh_writer_addrs: HashSet<SocketAddr> = writers
-                .iter()
-                .filter(|w| !w.draining.load(Ordering::Relaxed))
-                .filter(|w| w.generation == generation)
-                .map(|w| w.addr)
-                .collect();
-            let (fresh_ratio, fresh_missing_dc) =
-                Self::coverage_ratio(&desired_by_dc, &fresh_writer_addrs);
-            if !fresh_missing_dc.is_empty() {
-                warn!(
-                    previous_generation,
-                    generation,
-                    fresh_ratio = format_args!("{fresh_ratio:.3}"),
-                    missing_dc = ?fresh_missing_dc,
-                    "ME hardswap pending: fresh generation coverage incomplete"
-                );
-                return;
-            }
-        } else if !missing_dc.is_empty() {
-            warn!(
-                missing_dc = ?missing_dc,
-                // Keep stale writers alive when fresh coverage is incomplete.
-                "ME reinit coverage incomplete after map update; keeping stale writers"
-            );
-            return;
-        }
-
-        let desired_addrs: HashSet<SocketAddr> = desired_by_dc
-            .values()
-            .flat_map(|set| set.iter().copied())
-            .collect();
-
-        let stale_writer_ids: Vec<u64> = writers
-            .iter()
-            .filter(|w| !w.draining.load(Ordering::Relaxed))
-            .filter(|w| {
-                if hardswap {
-                    w.generation < generation
-                } else {
-                    !desired_addrs.contains(&w.addr)
-                }
-            })
-            .map(|w| w.id)
-            .collect();
-        drop(writers);
-
-        if stale_writer_ids.is_empty() {
-            debug!("ME map update completed with no stale writers");
-            return;
-        }
-
-        let drain_timeout = self.force_close_timeout();
-        let drain_timeout_secs = drain_timeout.map(|d| d.as_secs()).unwrap_or(0);
-        info!(
-            stale_writers = stale_writer_ids.len(),
-            previous_generation,
-            generation,
-            hardswap,
-            coverage_ratio = format_args!("{coverage_ratio:.3}"),
-            min_ratio = format_args!("{min_ratio:.3}"),
-            drain_timeout_secs,
-            "ME map update covered; draining stale writers"
-        );
-        self.stats.increment_pool_swap_total();
-        for writer_id in stale_writer_ids {
-            self.mark_writer_draining_with_timeout(writer_id, drain_timeout, !hardswap)
-                .await;
-        }
-    }
-
-    pub async fn update_proxy_maps(
-        &self,
-        new_v4: HashMap<i32, Vec<(IpAddr, u16)>>,
-        new_v6: Option<HashMap<i32, Vec<(IpAddr, u16)>>>,
-    ) -> bool {
-        let mut changed = false;
-        {
-            let mut guard = self.proxy_map_v4.write().await;
-            if !new_v4.is_empty() && *guard != new_v4 {
-                *guard = new_v4;
-                changed = true;
-            }
-        }
-        if let Some(v6) = new_v6 {
-            let mut guard = self.proxy_map_v6.write().await;
-            if !v6.is_empty() && *guard != v6 {
-                *guard = v6;
-                changed = true;
-            }
-        }
-        // Ensure negative DC entries mirror positives when absent (Telegram convention).
-        {
-            let mut guard = self.proxy_map_v4.write().await;
-            let keys: Vec<i32> = guard.keys().cloned().collect();
-            for k in keys.iter().cloned().filter(|k| *k > 0) {
-                if !guard.contains_key(&-k)
-                    && let Some(addrs) = guard.get(&k).cloned()
-                {
-                    guard.insert(-k, addrs);
-                }
-            }
-        }
-        {
-            let mut guard = self.proxy_map_v6.write().await;
-            let keys: Vec<i32> = guard.keys().cloned().collect();
-            for k in keys.iter().cloned().filter(|k| *k > 0) {
-                if !guard.contains_key(&-k)
-                    && let Some(addrs) = guard.get(&k).cloned()
-                {
-                    guard.insert(-k, addrs);
-                }
-            }
-        }
-        changed
-    }
-
-    pub async fn update_secret(self: &Arc<Self>, new_secret: Vec<u8>) -> bool {
-        if new_secret.len() < 32 {
-            warn!(len = new_secret.len(), "proxy-secret update ignored (too short)");
-            return false;
-        }
-        let mut guard = self.proxy_secret.write().await;
-        if *guard != new_secret {
-            *guard = new_secret;
-            drop(guard);
-            self.reconnect_all().await;
-            return true;
-        }
-        false
-    }
-
-    pub async fn reconnect_all(self: &Arc<Self>) {
-        let ws = self.writers.read().await.clone();
-        for w in ws {
-            if let Ok(()) = self.connect_one(w.addr, self.rng.as_ref()).await {
-                self.mark_writer_draining(w.id).await;
-                tokio::time::sleep(Duration::from_secs(2)).await;
-            }
-        }
-    }
-
     pub(super) async fn key_selector(&self) -> u32 {
         let secret = self.proxy_secret.read().await;
         if secret.len() >= 4 {
@@ -573,432 +335,13 @@ impl MePool {
         order
     }
 
-    async fn proxy_map_for_family(&self, family: IpFamily) -> HashMap<i32, Vec<(IpAddr, u16)>> {
+    pub(super) async fn proxy_map_for_family(
+        &self,
+        family: IpFamily,
+    ) -> HashMap<i32, Vec<(IpAddr, u16)>> {
         match family {
             IpFamily::V4 => self.proxy_map_v4.read().await.clone(),
             IpFamily::V6 => self.proxy_map_v6.read().await.clone(),
         }
     }
-
-    pub async fn init(self: &Arc<Self>, pool_size: usize, rng: &Arc<SecureRandom>) -> Result<()> {
-        let family_order = self.family_order();
-        let ks = self.key_selector().await;
-        info!(
-            me_servers = self.proxy_map_v4.read().await.len(),
-            pool_size,
-            key_selector = format_args!("0x{ks:08x}"),
-            secret_len = self.proxy_secret.read().await.len(),
-            "Initializing ME pool"
-        );
-
-        for family in family_order {
-            let map = self.proxy_map_for_family(family).await;
-            let dc_addrs: Vec<(i32, Vec<(IpAddr, u16)>)> = map
-                .iter()
-                .map(|(dc, addrs)| (*dc, addrs.clone()))
-                .collect();
-
-            // Ensure at least one connection per DC; run DCs in parallel.
-            let mut join = tokio::task::JoinSet::new();
-            let mut dc_failures = 0usize;
-            for (dc, addrs) in dc_addrs.iter().cloned() {
-                if addrs.is_empty() {
-                    continue;
-                }
-                let pool = Arc::clone(self);
-                let rng_clone = Arc::clone(rng);
-                join.spawn(async move {
-                    pool.connect_primary_for_dc(dc, addrs, rng_clone).await
-                });
-            }
-            while let Some(res) = join.join_next().await {
-                if let Ok(false) = res {
-                    dc_failures += 1;
-                }
-            }
-            if dc_failures > 2 {
-                return Err(ProxyError::Proxy("Too many ME DC init failures, falling back to direct".into()));
-            }
-
-        // Additional connections up to pool_size total (round-robin across DCs), staggered to de-phase lifecycles.
-        if self.me_warmup_stagger_enabled {
-            for (dc, addrs) in dc_addrs.iter() {
-                for (ip, port) in addrs {
-                    if self.connection_count() >= pool_size {
-                        break;
-                    }
-                    let addr = SocketAddr::new(*ip, *port);
-                    let jitter = rand::rng().random_range(0..=self.me_warmup_step_jitter.as_millis() as u64);
-                    let delay_ms = self.me_warmup_step_delay.as_millis() as u64 + jitter;
-                    tokio::time::sleep(Duration::from_millis(delay_ms)).await;
-                    if let Err(e) = self.connect_one(addr, rng.as_ref()).await {
-                        debug!(%addr, dc = %dc, error = %e, "Extra ME connect failed (staggered)");
-                    }
-                }
-            }
-        } else {
-            for (dc, addrs) in dc_addrs.iter() {
-                for (ip, port) in addrs {
-                    if self.connection_count() >= pool_size {
-                        break;
-                    }
-                    let addr = SocketAddr::new(*ip, *port);
-                    if let Err(e) = self.connect_one(addr, rng.as_ref()).await {
-                        debug!(%addr, dc = %dc, error = %e, "Extra ME connect failed");
-                    }
-                }
-                if self.connection_count() >= pool_size {
-                    break;
-                }
-            }
-        }
-
-            if !self.decision.effective_multipath && self.connection_count() > 0 {
-                break;
-            }
-        }
-
-        if self.writers.read().await.is_empty() {
-            return Err(ProxyError::Proxy("No ME connections".into()));
-        }
-        Ok(())
-    }
-
-    pub(crate) async fn connect_one(self: &Arc<Self>, addr: SocketAddr, rng: &SecureRandom) -> Result<()> {
-        let secret_len = self.proxy_secret.read().await.len();
-        if secret_len < 32 {
-            return Err(ProxyError::Proxy("proxy-secret too short for ME auth".into()));
-        }
-
-        let (stream, _connect_ms) = self.connect_tcp(addr).await?;
-        let hs = self.handshake_only(stream, addr, rng).await?;
-
-        let writer_id = self.next_writer_id.fetch_add(1, Ordering::Relaxed);
-        let generation = self.current_generation();
-        let cancel = CancellationToken::new();
-        let degraded = Arc::new(AtomicBool::new(false));
-        let draining = Arc::new(AtomicBool::new(false));
-        let draining_started_at_epoch_secs = Arc::new(AtomicU64::new(0));
-        let allow_drain_fallback = Arc::new(AtomicBool::new(false));
-        let (tx, mut rx) = mpsc::channel::<WriterCommand>(4096);
-        let mut rpc_writer = RpcWriter {
-            writer: hs.wr,
-            key: hs.write_key,
-            iv: hs.write_iv,
-            seq_no: 0,
-            crc_mode: hs.crc_mode,
-        };
-        let cancel_wr = cancel.clone();
-        tokio::spawn(async move {
-            loop {
-                tokio::select! {
-                    cmd = rx.recv() => {
-                        match cmd {
-                            Some(WriterCommand::Data(payload)) => {
-                                if rpc_writer.send(&payload).await.is_err() { break; }
-                            }
-                            Some(WriterCommand::DataAndFlush(payload)) => {
-                                if rpc_writer.send_and_flush(&payload).await.is_err() { break; }
-                            }
-                            Some(WriterCommand::Close) | None => break,
-                        }
-                    }
-                    _ = cancel_wr.cancelled() => break,
-                }
-            }
-        });
-        let writer = MeWriter {
-            id: writer_id,
-            addr,
-            generation,
-            tx: tx.clone(),
-            cancel: cancel.clone(),
-            degraded: degraded.clone(),
-            draining: draining.clone(),
-            draining_started_at_epoch_secs: draining_started_at_epoch_secs.clone(),
-            allow_drain_fallback: allow_drain_fallback.clone(),
-        };
-        self.writers.write().await.push(writer.clone());
-        self.conn_count.fetch_add(1, Ordering::Relaxed);
-        self.writer_available.notify_one();
-
-        let reg = self.registry.clone();
-        let writers_arc = self.writers_arc();
-        let ping_tracker = self.ping_tracker.clone();
-        let ping_tracker_reader = ping_tracker.clone();
-        let rtt_stats = self.rtt_stats.clone();
-        let stats_reader = self.stats.clone();
-        let stats_ping = self.stats.clone();
-        let pool = Arc::downgrade(self);
-        let cancel_ping = cancel.clone();
-        let tx_ping = tx.clone();
-        let ping_tracker_ping = ping_tracker.clone();
-        let cleanup_done = Arc::new(AtomicBool::new(false));
-        let cleanup_for_reader = cleanup_done.clone();
-        let cleanup_for_ping = cleanup_done.clone();
-        let keepalive_enabled = self.me_keepalive_enabled;
-        let keepalive_interval = self.me_keepalive_interval;
-        let keepalive_jitter = self.me_keepalive_jitter;
-        let cancel_reader_token = cancel.clone();
-        let cancel_ping_token = cancel_ping.clone();
-
-        tokio::spawn(async move {
-            let res = reader_loop(
-                hs.rd,
-                hs.read_key,
-                hs.read_iv,
-                hs.crc_mode,
-                reg.clone(),
-                BytesMut::new(),
-                BytesMut::new(),
-                tx.clone(),
-                ping_tracker_reader,
-                rtt_stats.clone(),
-                stats_reader,
-                writer_id,
-                degraded.clone(),
-                cancel_reader_token.clone(),
-            )
-            .await;
-            if let Some(pool) = pool.upgrade()
-                && cleanup_for_reader
-                    .compare_exchange(false, true, Ordering::AcqRel, Ordering::Relaxed)
-                    .is_ok()
-            {
-                pool.remove_writer_and_close_clients(writer_id).await;
-            }
-            if let Err(e) = res {
-                warn!(error = %e, "ME reader ended");
-            }
-            let mut ws = writers_arc.write().await;
-            ws.retain(|w| w.id != writer_id);
-            info!(remaining = ws.len(), "Dead ME writer removed from pool");
-        });
-
-        let pool_ping = Arc::downgrade(self);
-        tokio::spawn(async move {
-            let mut ping_id: i64 = rand::random::<i64>();
-            // Per-writer jittered start to avoid phase sync.
-            let startup_jitter = if keepalive_enabled {
-                let jitter_cap_ms = keepalive_interval.as_millis() / 2;
-                let effective_jitter_ms = keepalive_jitter.as_millis().min(jitter_cap_ms).max(1);
-                Duration::from_millis(rand::rng().random_range(0..=effective_jitter_ms as u64))
-            } else {
-                let jitter = rand::rng()
-                    .random_range(-ME_ACTIVE_PING_JITTER_SECS..=ME_ACTIVE_PING_JITTER_SECS);
-                let wait = (ME_ACTIVE_PING_SECS as i64 + jitter).max(5) as u64;
-                Duration::from_secs(wait)
-            };
-            tokio::select! {
-                _ = cancel_ping_token.cancelled() => return,
-                _ = tokio::time::sleep(startup_jitter) => {}
-            }
-            loop {
-                let wait = if keepalive_enabled {
-                    let jitter_cap_ms = keepalive_interval.as_millis() / 2;
-                    let effective_jitter_ms = keepalive_jitter.as_millis().min(jitter_cap_ms).max(1);
-                    keepalive_interval
-                        + Duration::from_millis(
-                            rand::rng().random_range(0..=effective_jitter_ms as u64)
-                        )
-                } else {
-                    let jitter = rand::rng()
-                        .random_range(-ME_ACTIVE_PING_JITTER_SECS..=ME_ACTIVE_PING_JITTER_SECS);
-                    let secs = (ME_ACTIVE_PING_SECS as i64 + jitter).max(5) as u64;
-                    Duration::from_secs(secs)
-                };
-                tokio::select! {
-                    _ = cancel_ping_token.cancelled() => {
-                        break;
-                    }
-                    _ = tokio::time::sleep(wait) => {}
-                }
-                let sent_id = ping_id;
-                let mut p = Vec::with_capacity(12);
-                p.extend_from_slice(&RPC_PING_U32.to_le_bytes());
-                p.extend_from_slice(&sent_id.to_le_bytes());
-                {
-                    let mut tracker = ping_tracker_ping.lock().await;
-                    let before = tracker.len();
-                    tracker.retain(|_, (ts, _)| ts.elapsed() < Duration::from_secs(120));
-                    let expired = before.saturating_sub(tracker.len());
-                    if expired > 0 {
-                        stats_ping.increment_me_keepalive_timeout_by(expired as u64);
-                    }
-                    tracker.insert(sent_id, (std::time::Instant::now(), writer_id));
-                }
-                ping_id = ping_id.wrapping_add(1);
-                stats_ping.increment_me_keepalive_sent();
-                if tx_ping.send(WriterCommand::DataAndFlush(p)).await.is_err() {
-                    stats_ping.increment_me_keepalive_failed();
-                    debug!("ME ping failed, removing dead writer");
-                    cancel_ping.cancel();
-                    if let Some(pool) = pool_ping.upgrade()
-                        && cleanup_for_ping
-                            .compare_exchange(false, true, Ordering::AcqRel, Ordering::Relaxed)
-                            .is_ok()
-                    {
-                        pool.remove_writer_and_close_clients(writer_id).await;
-                    }
-                    break;
-                }
-            }
-        });
-
-        Ok(())
-    }
-
-    async fn connect_primary_for_dc(
-        self: Arc<Self>,
-        dc: i32,
-        mut addrs: Vec<(IpAddr, u16)>,
-        rng: Arc<SecureRandom>,
-    ) -> bool {
-        if addrs.is_empty() {
-            return false;
-        }
-        addrs.shuffle(&mut rand::rng());
-        for (ip, port) in addrs {
-            let addr = SocketAddr::new(ip, port);
-            match self.connect_one(addr, rng.as_ref()).await {
-                Ok(()) => {
-                    info!(%addr, dc = %dc, "ME connected");
-                    return true;
-                }
-                Err(e) => warn!(%addr, dc = %dc, error = %e, "ME connect failed, trying next"),
-            }
-        }
-        warn!(dc = %dc, "All ME servers for DC failed at init");
-        false
-    }
-
-    pub(crate) async fn remove_writer_and_close_clients(self: &Arc<Self>, writer_id: u64) {
-        let conns = self.remove_writer_only(writer_id).await;
-        for bound in conns {
-            let _ = self.registry.route(bound.conn_id, super::MeResponse::Close).await;
-            let _ = self.registry.unregister(bound.conn_id).await;
-        }
-    }
-
-    async fn remove_writer_only(&self, writer_id: u64) -> Vec<BoundConn> {
-        let mut close_tx: Option<mpsc::Sender<WriterCommand>> = None;
-        {
-            let mut ws = self.writers.write().await;
-            if let Some(pos) = ws.iter().position(|w| w.id == writer_id) {
-                let w = ws.remove(pos);
-                if w.draining.load(Ordering::Relaxed) {
-                    self.stats.decrement_pool_drain_active();
-                }
-                w.cancel.cancel();
-                close_tx = Some(w.tx.clone());
-                self.conn_count.fetch_sub(1, Ordering::Relaxed);
-            }
-        }
-        if let Some(tx) = close_tx {
-            let _ = tx.send(WriterCommand::Close).await;
-        }
-        self.rtt_stats.lock().await.remove(&writer_id);
-        self.registry.writer_lost(writer_id).await
-    }
-
-    pub(crate) async fn mark_writer_draining_with_timeout(
-        self: &Arc<Self>,
-        writer_id: u64,
-        timeout: Option<Duration>,
-        allow_drain_fallback: bool,
-    ) {
-        let timeout = timeout.filter(|d| !d.is_zero());
-        let found = {
-            let mut ws = self.writers.write().await;
-            if let Some(w) = ws.iter_mut().find(|w| w.id == writer_id) {
-                let already_draining = w.draining.swap(true, Ordering::Relaxed);
-                w.allow_drain_fallback
-                    .store(allow_drain_fallback, Ordering::Relaxed);
-                w.draining_started_at_epoch_secs
-                    .store(Self::now_epoch_secs(), Ordering::Relaxed);
-                if !already_draining {
-                    self.stats.increment_pool_drain_active();
-                }
-                w.draining.store(true, Ordering::Relaxed);
-                true
-            } else {
-                false
-            }
-        };
-
-        if !found {
-            return;
-        }
-
-        let timeout_secs = timeout.map(|d| d.as_secs()).unwrap_or(0);
-        debug!(
-            writer_id,
-            timeout_secs,
-            allow_drain_fallback,
-            "ME writer marked draining"
-        );
-
-        let pool = Arc::downgrade(self);
-        tokio::spawn(async move {
-            let deadline = timeout.map(|t| Instant::now() + t);
-            while let Some(p) = pool.upgrade() {
-                if let Some(deadline_at) = deadline
-                    && Instant::now() >= deadline_at
-                {
-                    warn!(writer_id, "Drain timeout, force-closing");
-                    p.stats.increment_pool_force_close_total();
-                    let _ = p.remove_writer_and_close_clients(writer_id).await;
-                    break;
-                }
-                if p.registry.is_writer_empty(writer_id).await {
-                    let _ = p.remove_writer_only(writer_id).await;
-                    break;
-                }
-                tokio::time::sleep(Duration::from_secs(1)).await;
-            }
-        });
-    }
-
-    pub(crate) async fn mark_writer_draining(self: &Arc<Self>, writer_id: u64) {
-        self.mark_writer_draining_with_timeout(writer_id, Some(Duration::from_secs(300)), false)
-            .await;
-    }
-
-    pub(super) fn writer_accepts_new_binding(&self, writer: &MeWriter) -> bool {
-        if !writer.draining.load(Ordering::Relaxed) {
-            return true;
-        }
-        if !writer.allow_drain_fallback.load(Ordering::Relaxed) {
-            return false;
-        }
-
-        let ttl_secs = self.me_pool_drain_ttl_secs.load(Ordering::Relaxed);
-        if ttl_secs == 0 {
-            return true;
-        }
-
-        let started = writer.draining_started_at_epoch_secs.load(Ordering::Relaxed);
-        if started == 0 {
-            return false;
-        }
-
-        Self::now_epoch_secs().saturating_sub(started) <= ttl_secs
-    }
-
-}
-
-#[allow(dead_code)]
-fn hex_dump(data: &[u8]) -> String {
-    const MAX: usize = 64;
-    let mut out = String::with_capacity(data.len() * 2 + 3);
-    for (i, b) in data.iter().take(MAX).enumerate() {
-        if i > 0 {
-            out.push(' ');
-        }
-        out.push_str(&format!("{b:02x}"));
-    }
-    if data.len() > MAX {
-        out.push_str(" …");
-    }
-    out
 }

src/transport/middle_proxy/pool_config.rs

@@ -0,0 +1,81 @@
+use std::collections::HashMap;
+use std::net::IpAddr;
+use std::sync::Arc;
+use std::time::Duration;
+
+use tracing::warn;
+
+use super::pool::MePool;
+
+impl MePool {
+    pub async fn update_proxy_maps(
+        &self,
+        new_v4: HashMap<i32, Vec<(IpAddr, u16)>>,
+        new_v6: Option<HashMap<i32, Vec<(IpAddr, u16)>>>,
+    ) -> bool {
+        let mut changed = false;
+        {
+            let mut guard = self.proxy_map_v4.write().await;
+            if !new_v4.is_empty() && *guard != new_v4 {
+                *guard = new_v4;
+                changed = true;
+            }
+        }
+        if let Some(v6) = new_v6 {
+            let mut guard = self.proxy_map_v6.write().await;
+            if !v6.is_empty() && *guard != v6 {
+                *guard = v6;
+                changed = true;
+            }
+        }
+        // Ensure negative DC entries mirror positives when absent (Telegram convention).
+        {
+            let mut guard = self.proxy_map_v4.write().await;
+            let keys: Vec<i32> = guard.keys().cloned().collect();
+            for k in keys.iter().cloned().filter(|k| *k > 0) {
+                if !guard.contains_key(&-k)
+                    && let Some(addrs) = guard.get(&k).cloned()
+                {
+                    guard.insert(-k, addrs);
+                }
+            }
+        }
+        {
+            let mut guard = self.proxy_map_v6.write().await;
+            let keys: Vec<i32> = guard.keys().cloned().collect();
+            for k in keys.iter().cloned().filter(|k| *k > 0) {
+                if !guard.contains_key(&-k)
+                    && let Some(addrs) = guard.get(&k).cloned()
+                {
+                    guard.insert(-k, addrs);
+                }
+            }
+        }
+        changed
+    }
+
+    pub async fn update_secret(self: &Arc<Self>, new_secret: Vec<u8>) -> bool {
+        if new_secret.len() < 32 {
+            warn!(len = new_secret.len(), "proxy-secret update ignored (too short)");
+            return false;
+        }
+        let mut guard = self.proxy_secret.write().await;
+        if *guard != new_secret {
+            *guard = new_secret;
+            drop(guard);
+            self.reconnect_all().await;
+            return true;
+        }
+        false
+    }
+
+    pub async fn reconnect_all(self: &Arc<Self>) {
+        let ws = self.writers.read().await.clone();
+        for w in ws {
+            if let Ok(()) = self.connect_one(w.addr, self.rng.as_ref()).await {
+                self.mark_writer_draining(w.id).await;
+                tokio::time::sleep(Duration::from_secs(2)).await;
+            }
+        }
+    }
+}

src/transport/middle_proxy/pool_init.rs

@@ -0,0 +1,201 @@
+use std::collections::{HashMap, HashSet};
+use std::net::{IpAddr, SocketAddr};
+use std::sync::Arc;
+
+use rand::Rng;
+use rand::seq::SliceRandom;
+use tracing::{debug, info, warn};
+
+use crate::crypto::SecureRandom;
+use crate::error::{ProxyError, Result};
+
+use super::pool::MePool;
+
+impl MePool {
+    pub async fn init(self: &Arc<Self>, pool_size: usize, rng: &Arc<SecureRandom>) -> Result<()> {
+        let family_order = self.family_order();
+        let ks = self.key_selector().await;
+        info!(
+            me_servers = self.proxy_map_v4.read().await.len(),
+            pool_size,
+            key_selector = format_args!("0x{ks:08x}"),
+            secret_len = self.proxy_secret.read().await.len(),
+            "Initializing ME pool"
+        );
+
+        for family in family_order {
+            let map = self.proxy_map_for_family(family).await;
+            let mut grouped_dc_addrs: HashMap<i32, Vec<(IpAddr, u16)>> = HashMap::new();
+            for (dc, addrs) in map {
+                if addrs.is_empty() {
+                    continue;
+                }
+                grouped_dc_addrs.entry(dc.abs()).or_default().extend(addrs);
+            }
+            let mut dc_addrs: Vec<(i32, Vec<(IpAddr, u16)>)> = grouped_dc_addrs
+                .into_iter()
+                .map(|(dc, mut addrs)| {
+                    addrs.sort_unstable();
+                    addrs.dedup();
+                    (dc, addrs)
+                })
+                .collect();
+            dc_addrs.sort_unstable_by_key(|(dc, _)| *dc);
+
+            // Ensure at least one live writer per DC group; run missing DCs in parallel.
+            let mut join = tokio::task::JoinSet::new();
+            for (dc, addrs) in dc_addrs.iter().cloned() {
+                if addrs.is_empty() {
+                    continue;
+                }
+                let endpoints: HashSet<SocketAddr> = addrs
+                    .iter()
+                    .map(|(ip, port)| SocketAddr::new(*ip, *port))
+                    .collect();
+                if self.active_writer_count_for_endpoints(&endpoints).await > 0 {
+                    continue;
+                }
+                let pool = Arc::clone(self);
+                let rng_clone = Arc::clone(rng);
+                join.spawn(async move { pool.connect_primary_for_dc(dc, addrs, rng_clone).await });
+            }
+            while join.join_next().await.is_some() {}
+
+            let mut missing_dcs = Vec::new();
+            for (dc, addrs) in &dc_addrs {
+                let endpoints: HashSet<SocketAddr> = addrs
+                    .iter()
+                    .map(|(ip, port)| SocketAddr::new(*ip, *port))
+                    .collect();
+                if self.active_writer_count_for_endpoints(&endpoints).await == 0 {
+                    missing_dcs.push(*dc);
+                }
+            }
+            if !missing_dcs.is_empty() {
+                return Err(ProxyError::Proxy(format!(
+                    "ME init incomplete: no live writers for DC groups {missing_dcs:?}"
+                )));
+            }
+
+            // Warm reserve writers asynchronously so startup does not block after first working pool is ready.
+            let pool = Arc::clone(self);
+            let rng_clone = Arc::clone(rng);
+            let dc_addrs_bg = dc_addrs.clone();
+            tokio::spawn(async move {
+                if pool.me_warmup_stagger_enabled {
+                    for (dc, addrs) in &dc_addrs_bg {
+                        for (ip, port) in addrs {
+                            if pool.connection_count() >= pool_size {
+                                break;
+                            }
+                            let addr = SocketAddr::new(*ip, *port);
+                            let jitter = rand::rng()
+                                .random_range(0..=pool.me_warmup_step_jitter.as_millis() as u64);
+                            let delay_ms = pool.me_warmup_step_delay.as_millis() as u64 + jitter;
+                            tokio::time::sleep(std::time::Duration::from_millis(delay_ms)).await;
+                            if let Err(e) = pool.connect_one(addr, rng_clone.as_ref()).await {
+                                debug!(%addr, dc = %dc, error = %e, "Extra ME connect failed (staggered)");
+                            }
+                        }
+                    }
+                } else {
+                    for (dc, addrs) in &dc_addrs_bg {
+                        for (ip, port) in addrs {
+                            if pool.connection_count() >= pool_size {
+                                break;
+                            }
+                            let addr = SocketAddr::new(*ip, *port);
+                            if let Err(e) = pool.connect_one(addr, rng_clone.as_ref()).await {
+                                debug!(%addr, dc = %dc, error = %e, "Extra ME connect failed");
+                            }
+                        }
+                        if pool.connection_count() >= pool_size {
+                            break;
+                        }
+                    }
+                }
+                debug!(
+                    target_pool_size = pool_size,
+                    current_pool_size = pool.connection_count(),
+                    "Background ME reserve warmup finished"
+                );
+            });
+
+            if !self.decision.effective_multipath && self.connection_count() > 0 {
+                break;
+            }
+        }
+
+        if self.writers.read().await.is_empty() {
+            return Err(ProxyError::Proxy("No ME connections".into()));
+        }
+        info!(
+            active_writers = self.connection_count(),
+            "ME primary pool ready; reserve warmup continues in background"
+        );
+        Ok(())
+    }
+
+    async fn connect_primary_for_dc(
+        self: Arc<Self>,
+        dc: i32,
+        mut addrs: Vec<(IpAddr, u16)>,
+        rng: Arc<SecureRandom>,
+    ) -> bool {
+        if addrs.is_empty() {
+            return false;
+        }
+        addrs.shuffle(&mut rand::rng());
+        if addrs.len() > 1 {
+            let concurrency = 2usize;
+            let mut join = tokio::task::JoinSet::new();
+            let mut next_idx = 0usize;
+
+            while next_idx < addrs.len() || !join.is_empty() {
+                while next_idx < addrs.len() && join.len() < concurrency {
+                    let (ip, port) = addrs[next_idx];
+                    next_idx += 1;
+                    let addr = SocketAddr::new(ip, port);
+                    let pool = Arc::clone(&self);
+                    let rng_clone = Arc::clone(&rng);
+                    join.spawn(async move {
+                        (addr, pool.connect_one(addr, rng_clone.as_ref()).await)
+                    });
+                }
+
+                let Some(res) = join.join_next().await else {
+                    break;
+                };
+                match res {
+                    Ok((addr, Ok(()))) => {
+                        info!(%addr, dc = %dc, "ME connected");
+                        join.abort_all();
+                        while join.join_next().await.is_some() {}
+                        return true;
+                    }
+                    Ok((addr, Err(e))) => {
+                        warn!(%addr, dc = %dc, error = %e, "ME connect failed, trying next");
+                    }
+                    Err(e) => {
+                        warn!(dc = %dc, error = %e, "ME connect task failed");
+                    }
+                }
+            }
+            warn!(dc = %dc, "All ME servers for DC failed at init");
+            return false;
+        }
+
+        for (ip, port) in addrs {
+            let addr = SocketAddr::new(ip, port);
+            match self.connect_one(addr, rng.as_ref()).await {
+                Ok(()) => {
+                    info!(%addr, dc = %dc, "ME connected");
+                    return true;
+                }
+                Err(e) => warn!(%addr, dc = %dc, error = %e, "ME connect failed, trying next"),
+            }
+        }
+        warn!(dc = %dc, "All ME servers for DC failed at init");
+        false
+    }
+}

src/transport/middle_proxy/pool_nat.rs

@@ -1,18 +1,31 @@
+use std::collections::HashMap;
 use std::net::{IpAddr, Ipv4Addr};
 use std::time::Duration;
 
-use tracing::{info, warn};
+use tokio::task::JoinSet;
+use tokio::time::timeout;
+use tracing::{debug, info, warn};
 
 use crate::error::{ProxyError, Result};
 use crate::network::probe::is_bogon;
-use crate::network::stun::{stun_probe_dual, IpFamily, StunProbeResult};
+use crate::network::stun::{stun_probe_dual, stun_probe_family_with_bind, IpFamily};
 
 use super::MePool;
 use std::time::Instant;
 
+const STUN_BATCH_TIMEOUT: Duration = Duration::from_secs(5);
+
 #[allow(dead_code)]
 pub async fn stun_probe(stun_addr: Option<String>) -> Result<crate::network::stun::DualStunResult> {
-    let stun_addr = stun_addr.unwrap_or_else(|| "stun.l.google.com:19302".to_string());
+    let stun_addr = stun_addr.unwrap_or_else(|| {
+        crate::config::defaults::default_stun_servers()
+            .into_iter()
+            .next()
+            .unwrap_or_default()
+    });
+    if stun_addr.is_empty() {
+        return Err(ProxyError::Proxy("STUN server is not configured".to_string()));
+    }
     stun_probe_dual(&stun_addr).await
 }
 
@@ -22,6 +35,101 @@ pub async fn detect_public_ip() -> Option<IpAddr> {
 }
 
 impl MePool {
+    fn configured_stun_servers(&self) -> Vec<String> {
+        if !self.nat_stun_servers.is_empty() {
+            return self.nat_stun_servers.clone();
+        }
+        if let Some(s) = &self.nat_stun
+            && !s.trim().is_empty()
+        {
+            return vec![s.clone()];
+        }
+        Vec::new()
+    }
+
+    async fn probe_stun_batch_for_family(
+        &self,
+        servers: &[String],
+        family: IpFamily,
+        attempt: u8,
+        bind_ip: Option<IpAddr>,
+    ) -> (Vec<String>, Option<std::net::SocketAddr>) {
+        let mut join_set = JoinSet::new();
+        let mut next_idx = 0usize;
+        let mut live_servers = Vec::new();
+        let mut best_by_ip: HashMap<IpAddr, (usize, std::net::SocketAddr)> = HashMap::new();
+        let concurrency = self.nat_probe_concurrency.max(1);
+
+        while next_idx < servers.len() || !join_set.is_empty() {
+            while next_idx < servers.len() && join_set.len() < concurrency {
+                let stun_addr = servers[next_idx].clone();
+                next_idx += 1;
+                join_set.spawn(async move {
+                    let res = timeout(
+                        STUN_BATCH_TIMEOUT,
+                        stun_probe_family_with_bind(&stun_addr, family, bind_ip),
+                    )
+                    .await;
+                    (stun_addr, res)
+                });
+            }
+
+            let Some(task) = join_set.join_next().await else {
+                break;
+            };
+
+            match task {
+                Ok((stun_addr, Ok(Ok(picked)))) => {
+                    if let Some(result) = picked {
+                        live_servers.push(stun_addr.clone());
+                        let entry = best_by_ip
+                            .entry(result.reflected_addr.ip())
+                            .or_insert((0, result.reflected_addr));
+                        entry.0 += 1;
+                        debug!(
+                            local = %result.local_addr,
+                            reflected = %result.reflected_addr,
+                            family = ?family,
+                            stun = %stun_addr,
+                            "NAT probe: reflected address"
+                        );
+                    }
+                }
+                Ok((stun_addr, Ok(Err(e)))) => {
+                    debug!(
+                        error = %e,
+                        stun = %stun_addr,
+                        attempt = attempt + 1,
+                        "NAT probe failed, trying next server"
+                    );
+                }
+                Ok((stun_addr, Err(_))) => {
+                    debug!(
+                        stun = %stun_addr,
+                        attempt = attempt + 1,
+                        "NAT probe timeout, trying next server"
+                    );
+                }
+                Err(e) => {
+                    debug!(
+                        error = %e,
+                        attempt = attempt + 1,
+                        "NAT probe task join failed"
+                    );
+                }
+            }
+        }
+
+        live_servers.sort_unstable();
+        live_servers.dedup();
+        let best_reflected = best_by_ip
+            .into_values()
+            .max_by_key(|(count, _)| *count)
+            .map(|(_, addr)| addr);
+
+        (live_servers, best_reflected)
+    }
+
     pub(super) fn translate_ip_for_nat(&self, ip: IpAddr) -> IpAddr {
         let nat_ip = self
             .nat_ip_cfg
@@ -99,10 +207,21 @@ impl MePool {
     pub(super) async fn maybe_reflect_public_addr(
         &self,
         family: IpFamily,
+        bind_ip: Option<IpAddr>,
     ) -> Option<std::net::SocketAddr> {
         const STUN_CACHE_TTL: Duration = Duration::from_secs(600);
+        let use_shared_cache = bind_ip.is_none();
+        if !use_shared_cache {
+            match (family, bind_ip) {
+                (IpFamily::V4, Some(IpAddr::V4(_)))
+                | (IpFamily::V6, Some(IpAddr::V6(_)))
+                | (_, None) => {}
+                _ => return None,
+            }
+        }
         // Backoff window
-        if let Some(until) = *self.stun_backoff_until.read().await
+        if use_shared_cache
+            && let Some(until) = *self.stun_backoff_until.read().await
             && Instant::now() < until
         {
             if let Ok(cache) = self.nat_reflection_cache.try_lock() {
@@ -115,7 +234,9 @@ impl MePool {
             return None;
         }
 
-        if let Ok(mut cache) = self.nat_reflection_cache.try_lock() {
+        if use_shared_cache
+            && let Ok(mut cache) = self.nat_reflection_cache.try_lock()
+        {
             let slot = match family {
                 IpFamily::V4 => &mut cache.v4,
                 IpFamily::V6 => &mut cache.v6,
@@ -127,42 +248,64 @@ impl MePool {
             }
         }
 
-        let attempt = self.nat_probe_attempts.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
-        let servers = if !self.nat_stun_servers.is_empty() {
-            self.nat_stun_servers.clone()
-        } else if let Some(s) = &self.nat_stun {
-            vec![s.clone()]
+        let attempt = if use_shared_cache {
+            self.nat_probe_attempts.fetch_add(1, std::sync::atomic::Ordering::Relaxed)
         } else {
-            vec!["stun.l.google.com:19302".to_string()]
+            0
+        };
+        let configured_servers = self.configured_stun_servers();
+        let live_snapshot = self.nat_stun_live_servers.read().await.clone();
+        let primary_servers = if live_snapshot.is_empty() {
+            configured_servers.clone()
+        } else {
+            live_snapshot
         };
 
-        for stun_addr in servers {
-            match stun_probe_dual(&stun_addr).await {
-                Ok(res) => {
-                    let picked: Option<StunProbeResult> = match family {
-                        IpFamily::V4 => res.v4,
-                        IpFamily::V6 => res.v6,
-                    };
-                    if let Some(result) = picked {
-                        info!(local = %result.local_addr, reflected = %result.reflected_addr, family = ?family, stun = %stun_addr, "NAT probe: reflected address");
-                        self.nat_probe_attempts.store(0, std::sync::atomic::Ordering::Relaxed);
-                        if let Ok(mut cache) = self.nat_reflection_cache.try_lock() {
-                            let slot = match family {
-                                IpFamily::V4 => &mut cache.v4,
-                                IpFamily::V6 => &mut cache.v6,
-                            };
-                            *slot = Some((Instant::now(), result.reflected_addr));
-                        }
-                        return Some(result.reflected_addr);
-                    }
-                }
-                Err(e) => {
-                    warn!(error = %e, stun = %stun_addr, attempt = attempt + 1, "NAT probe failed, trying next server");
-                }
-            }
+        let (mut live_servers, mut selected_reflected) = self
+            .probe_stun_batch_for_family(&primary_servers, family, attempt, bind_ip)
+            .await;
+
+        if selected_reflected.is_none() && !configured_servers.is_empty() && primary_servers != configured_servers {
+            let (rediscovered_live, rediscovered_reflected) = self
+                .probe_stun_batch_for_family(&configured_servers, family, attempt, bind_ip)
+                .await;
+            live_servers = rediscovered_live;
+            selected_reflected = rediscovered_reflected;
+        }
+
+        let live_server_count = live_servers.len();
+        if !live_servers.is_empty() {
+            *self.nat_stun_live_servers.write().await = live_servers;
+        } else {
+            self.nat_stun_live_servers.write().await.clear();
+        }
+
+        if let Some(reflected_addr) = selected_reflected {
+            if use_shared_cache {
+                self.nat_probe_attempts.store(0, std::sync::atomic::Ordering::Relaxed);
+            }
+            info!(
+                family = ?family,
+                live_servers = live_server_count,
+                "STUN-Quorum reached, IP: {}",
+                reflected_addr.ip()
+            );
+            if use_shared_cache
+                && let Ok(mut cache) = self.nat_reflection_cache.try_lock()
+            {
+                let slot = match family {
+                    IpFamily::V4 => &mut cache.v4,
+                    IpFamily::V6 => &mut cache.v6,
+                };
+                *slot = Some((Instant::now(), reflected_addr));
+            }
+            return Some(reflected_addr);
+        }
+
+        if use_shared_cache {
+            let backoff = Duration::from_secs(60 * 2u64.pow((attempt as u32).min(6)));
+            *self.stun_backoff_until.write().await = Some(Instant::now() + backoff);
         }
-        let backoff = Duration::from_secs(60 * 2u64.pow((attempt as u32).min(6)));
-        *self.stun_backoff_until.write().await = Some(Instant::now() + backoff);
         None
     }
 }

src/transport/middle_proxy/pool_refill.rs

@@ -0,0 +1,159 @@
+use std::collections::HashSet;
+use std::net::SocketAddr;
+use std::sync::Arc;
+use std::sync::atomic::Ordering;
+
+use tracing::{debug, info, warn};
+
+use crate::crypto::SecureRandom;
+
+use super::pool::MePool;
+
+impl MePool {
+    pub(super) async fn connect_endpoints_round_robin(
+        self: &Arc<Self>,
+        endpoints: &[SocketAddr],
+        rng: &SecureRandom,
+    ) -> bool {
+        if endpoints.is_empty() {
+            return false;
+        }
+        let start = (self.rr.fetch_add(1, Ordering::Relaxed) as usize) % endpoints.len();
+        for offset in 0..endpoints.len() {
+            let idx = (start + offset) % endpoints.len();
+            let addr = endpoints[idx];
+            match self.connect_one(addr, rng).await {
+                Ok(()) => return true,
+                Err(e) => debug!(%addr, error = %e, "ME connect failed during round-robin warmup"),
+            }
+        }
+        false
+    }
+
+    async fn endpoints_for_same_dc(&self, addr: SocketAddr) -> Vec<SocketAddr> {
+        let mut target_dc = HashSet::<i32>::new();
+        let mut endpoints = HashSet::<SocketAddr>::new();
+
+        if self.decision.ipv4_me {
+            let map = self.proxy_map_v4.read().await.clone();
+            for (dc, addrs) in &map {
+                if addrs
+                    .iter()
+                    .any(|(ip, port)| SocketAddr::new(*ip, *port) == addr)
+                {
+                    target_dc.insert(dc.abs());
+                }
+            }
+            for dc in &target_dc {
+                for key in [*dc, -*dc] {
+                    if let Some(addrs) = map.get(&key) {
+                        for (ip, port) in addrs {
+                            endpoints.insert(SocketAddr::new(*ip, *port));
+                        }
+                    }
+                }
+            }
+        }
+
+        if self.decision.ipv6_me {
+            let map = self.proxy_map_v6.read().await.clone();
+            for (dc, addrs) in &map {
+                if addrs
+                    .iter()
+                    .any(|(ip, port)| SocketAddr::new(*ip, *port) == addr)
+                {
+                    target_dc.insert(dc.abs());
+                }
+            }
+            for dc in &target_dc {
+                for key in [*dc, -*dc] {
+                    if let Some(addrs) = map.get(&key) {
+                        for (ip, port) in addrs {
+                            endpoints.insert(SocketAddr::new(*ip, *port));
+                        }
+                    }
+                }
+            }
+        }
+
+        let mut sorted: Vec<SocketAddr> = endpoints.into_iter().collect();
+        sorted.sort_unstable();
+        sorted
+    }
+
+    async fn refill_writer_after_loss(self: &Arc<Self>, addr: SocketAddr) -> bool {
+        let fast_retries = self.me_reconnect_fast_retry_count.max(1);
+
+        for attempt in 0..fast_retries {
+            self.stats.increment_me_reconnect_attempt();
+            match self.connect_one(addr, self.rng.as_ref()).await {
+                Ok(()) => {
+                    self.stats.increment_me_reconnect_success();
+                    self.stats.increment_me_writer_restored_same_endpoint_total();
+                    info!(
+                        %addr,
+                        attempt = attempt + 1,
+                        "ME writer restored on the same endpoint"
+                    );
+                    return true;
+                }
+                Err(e) => {
+                    debug!(
+                        %addr,
+                        attempt = attempt + 1,
+                        error = %e,
+                        "ME immediate same-endpoint reconnect failed"
+                    );
+                }
+            }
+        }
+
+        let dc_endpoints = self.endpoints_for_same_dc(addr).await;
+        if dc_endpoints.is_empty() {
+            self.stats.increment_me_refill_failed_total();
+            return false;
+        }
+
+        for attempt in 0..fast_retries {
+            self.stats.increment_me_reconnect_attempt();
+            if self
+                .connect_endpoints_round_robin(&dc_endpoints, self.rng.as_ref())
+                .await
+            {
+                self.stats.increment_me_reconnect_success();
+                self.stats.increment_me_writer_restored_fallback_total();
+                info!(
+                    %addr,
+                    attempt = attempt + 1,
+                    "ME writer restored via DC fallback endpoint"
+                );
+                return true;
+            }
+        }
+
+        self.stats.increment_me_refill_failed_total();
+        false
+    }
+
+    pub(crate) fn trigger_immediate_refill(self: &Arc<Self>, addr: SocketAddr) {
+        let pool = Arc::clone(self);
+        tokio::spawn(async move {
+            {
+                let mut guard = pool.refill_inflight.lock().await;
+                if !guard.insert(addr) {
+                    pool.stats.increment_me_refill_skipped_inflight_total();
+                    return;
+                }
+            }
+            pool.stats.increment_me_refill_triggered_total();
+
+            let restored = pool.refill_writer_after_loss(addr).await;
+            if !restored {
+                warn!(%addr, "ME immediate refill failed");
+            }
+
+            let mut guard = pool.refill_inflight.lock().await;
+            guard.remove(&addr);
+        });
+    }
+}

src/transport/middle_proxy/pool_reinit.rs

@@ -0,0 +1,383 @@
+use std::collections::{HashMap, HashSet};
+use std::net::SocketAddr;
+use std::sync::Arc;
+use std::sync::atomic::Ordering;
+use std::time::Duration;
+
+use rand::Rng;
+use rand::seq::SliceRandom;
+use tracing::{debug, info, warn};
+
+use crate::crypto::SecureRandom;
+
+use super::pool::MePool;
+
+impl MePool {
+    fn coverage_ratio(
+        desired_by_dc: &HashMap<i32, HashSet<SocketAddr>>,
+        active_writer_addrs: &HashSet<SocketAddr>,
+    ) -> (f32, Vec<i32>) {
+        if desired_by_dc.is_empty() {
+            return (1.0, Vec::new());
+        }
+
+        let mut missing_dc = Vec::<i32>::new();
+        let mut covered = 0usize;
+        for (dc, endpoints) in desired_by_dc {
+            if endpoints.is_empty() {
+                continue;
+            }
+            if endpoints
+                .iter()
+                .any(|addr| active_writer_addrs.contains(addr))
+            {
+                covered += 1;
+            } else {
+                missing_dc.push(*dc);
+            }
+        }
+
+        missing_dc.sort_unstable();
+        let total = desired_by_dc.len().max(1);
+        let ratio = (covered as f32) / (total as f32);
+        (ratio, missing_dc)
+    }
+
+    pub async fn reconcile_connections(self: &Arc<Self>, rng: &SecureRandom) {
+        let writers = self.writers.read().await;
+        let current: HashSet<SocketAddr> = writers
+            .iter()
+            .filter(|w| !w.draining.load(Ordering::Relaxed))
+            .map(|w| w.addr)
+            .collect();
+        drop(writers);
+
+        for family in self.family_order() {
+            let map = self.proxy_map_for_family(family).await;
+            for (_dc, addrs) in &map {
+                let dc_addrs: Vec<SocketAddr> = addrs
+                    .iter()
+                    .map(|(ip, port)| SocketAddr::new(*ip, *port))
+                    .collect();
+                if !dc_addrs.iter().any(|a| current.contains(a)) {
+                    let mut shuffled = dc_addrs.clone();
+                    shuffled.shuffle(&mut rand::rng());
+                    for addr in shuffled {
+                        if self.connect_one(addr, rng).await.is_ok() {
+                            break;
+                        }
+                    }
+                }
+            }
+            if !self.decision.effective_multipath && !current.is_empty() {
+                break;
+            }
+        }
+    }
+
+    async fn desired_dc_endpoints(&self) -> HashMap<i32, HashSet<SocketAddr>> {
+        let mut out: HashMap<i32, HashSet<SocketAddr>> = HashMap::new();
+
+        if self.decision.ipv4_me {
+            let map_v4 = self.proxy_map_v4.read().await.clone();
+            for (dc, addrs) in map_v4 {
+                let entry = out.entry(dc.abs()).or_default();
+                for (ip, port) in addrs {
+                    entry.insert(SocketAddr::new(ip, port));
+                }
+            }
+        }
+
+        if self.decision.ipv6_me {
+            let map_v6 = self.proxy_map_v6.read().await.clone();
+            for (dc, addrs) in map_v6 {
+                let entry = out.entry(dc.abs()).or_default();
+                for (ip, port) in addrs {
+                    entry.insert(SocketAddr::new(ip, port));
+                }
+            }
+        }
+
+        out
+    }
+
+    pub(super) fn required_writers_for_dc(endpoint_count: usize) -> usize {
+        endpoint_count.max(3)
+    }
+
+    fn hardswap_warmup_connect_delay_ms(&self) -> u64 {
+        let min_ms = self.me_hardswap_warmup_delay_min_ms.load(Ordering::Relaxed);
+        let max_ms = self.me_hardswap_warmup_delay_max_ms.load(Ordering::Relaxed);
+        let (min_ms, max_ms) = if min_ms <= max_ms {
+            (min_ms, max_ms)
+        } else {
+            (max_ms, min_ms)
+        };
+        if min_ms == max_ms {
+            return min_ms;
+        }
+        rand::rng().random_range(min_ms..=max_ms)
+    }
+
+    fn hardswap_warmup_backoff_ms(&self, pass_idx: usize) -> u64 {
+        let base_ms = self
+            .me_hardswap_warmup_pass_backoff_base_ms
+            .load(Ordering::Relaxed);
+        let cap_ms = (self.me_reconnect_backoff_cap.as_millis() as u64).max(base_ms);
+        let shift = (pass_idx as u32).min(20);
+        let scaled = base_ms.saturating_mul(1u64 << shift);
+        let core = scaled.min(cap_ms);
+        let jitter = (core / 2).max(1);
+        core.saturating_add(rand::rng().random_range(0..=jitter))
+    }
+
+    async fn fresh_writer_count_for_endpoints(
+        &self,
+        generation: u64,
+        endpoints: &HashSet<SocketAddr>,
+    ) -> usize {
+        let ws = self.writers.read().await;
+        ws.iter()
+            .filter(|w| !w.draining.load(Ordering::Relaxed))
+            .filter(|w| w.generation == generation)
+            .filter(|w| endpoints.contains(&w.addr))
+            .count()
+    }
+
+    pub(super) async fn active_writer_count_for_endpoints(
+        &self,
+        endpoints: &HashSet<SocketAddr>,
+    ) -> usize {
+        let ws = self.writers.read().await;
+        ws.iter()
+            .filter(|w| !w.draining.load(Ordering::Relaxed))
+            .filter(|w| endpoints.contains(&w.addr))
+            .count()
+    }
+
+    async fn warmup_generation_for_all_dcs(
+        self: &Arc<Self>,
+        rng: &SecureRandom,
+        generation: u64,
+        desired_by_dc: &HashMap<i32, HashSet<SocketAddr>>,
+    ) {
+        let extra_passes = self
+            .me_hardswap_warmup_extra_passes
+            .load(Ordering::Relaxed)
+            .min(10) as usize;
+        let total_passes = 1 + extra_passes;
+
+        for (dc, endpoints) in desired_by_dc {
+            if endpoints.is_empty() {
+                continue;
+            }
+
+            let mut endpoint_list: Vec<SocketAddr> = endpoints.iter().copied().collect();
+            endpoint_list.sort_unstable();
+            let required = Self::required_writers_for_dc(endpoint_list.len());
+            let mut completed = false;
+            let mut last_fresh_count = self
+                .fresh_writer_count_for_endpoints(generation, endpoints)
+                .await;
+
+            for pass_idx in 0..total_passes {
+                if last_fresh_count >= required {
+                    completed = true;
+                    break;
+                }
+
+                let missing = required.saturating_sub(last_fresh_count);
+                debug!(
+                    dc = *dc,
+                    pass = pass_idx + 1,
+                    total_passes,
+                    fresh_count = last_fresh_count,
+                    required,
+                    missing,
+                    endpoint_count = endpoint_list.len(),
+                    "ME hardswap warmup pass started"
+                );
+
+                for attempt_idx in 0..missing {
+                    let delay_ms = self.hardswap_warmup_connect_delay_ms();
+                    tokio::time::sleep(Duration::from_millis(delay_ms)).await;
+
+                    let connected = self.connect_endpoints_round_robin(&endpoint_list, rng).await;
+                    debug!(
+                        dc = *dc,
+                        pass = pass_idx + 1,
+                        total_passes,
+                        attempt = attempt_idx + 1,
+                        delay_ms,
+                        connected,
+                        "ME hardswap warmup connect attempt finished"
+                    );
+                }
+
+                last_fresh_count = self
+                    .fresh_writer_count_for_endpoints(generation, endpoints)
+                    .await;
+                if last_fresh_count >= required {
+                    completed = true;
+                    info!(
+                        dc = *dc,
+                        pass = pass_idx + 1,
+                        total_passes,
+                        fresh_count = last_fresh_count,
+                        required,
+                        "ME hardswap warmup floor reached for DC"
+                    );
+                    break;
+                }
+
+                if pass_idx + 1 < total_passes {
+                    let backoff_ms = self.hardswap_warmup_backoff_ms(pass_idx);
+                    debug!(
+                        dc = *dc,
+                        pass = pass_idx + 1,
+                        total_passes,
+                        fresh_count = last_fresh_count,
+                        required,
+                        backoff_ms,
+                        "ME hardswap warmup pass incomplete, delaying next pass"
+                    );
+                    tokio::time::sleep(Duration::from_millis(backoff_ms)).await;
+                }
+            }
+
+            if !completed {
+                warn!(
+                    dc = *dc,
+                    fresh_count = last_fresh_count,
+                    required,
+                    endpoint_count = endpoint_list.len(),
+                    total_passes,
+                    "ME warmup stopped: unable to reach required writer floor for DC"
+                );
+            }
+        }
+    }
+
+    pub async fn zero_downtime_reinit_after_map_change(self: &Arc<Self>, rng: &SecureRandom) {
+        let desired_by_dc = self.desired_dc_endpoints().await;
+        if desired_by_dc.is_empty() {
+            warn!("ME endpoint map is empty; skipping stale writer drain");
+            return;
+        }
+
+        let previous_generation = self.current_generation();
+        let generation = self.generation.fetch_add(1, Ordering::Relaxed) + 1;
+        let hardswap = self.hardswap.load(Ordering::Relaxed);
+
+        if hardswap {
+            self.warmup_generation_for_all_dcs(rng, generation, &desired_by_dc)
+                .await;
+        } else {
+            self.reconcile_connections(rng).await;
+        }
+
+        let writers = self.writers.read().await;
+        let active_writer_addrs: HashSet<SocketAddr> = writers
+            .iter()
+            .filter(|w| !w.draining.load(Ordering::Relaxed))
+            .map(|w| w.addr)
+            .collect();
+        let min_ratio = Self::permille_to_ratio(
+            self.me_pool_min_fresh_ratio_permille
+                .load(Ordering::Relaxed),
+        );
+        let (coverage_ratio, missing_dc) = Self::coverage_ratio(&desired_by_dc, &active_writer_addrs);
+        if !hardswap && coverage_ratio < min_ratio {
+            warn!(
+                previous_generation,
+                generation,
+                coverage_ratio = format_args!("{coverage_ratio:.3}"),
+                min_ratio = format_args!("{min_ratio:.3}"),
+                missing_dc = ?missing_dc,
+                "ME reinit coverage below threshold; keeping stale writers"
+            );
+            return;
+        }
+
+        if hardswap {
+            let mut fresh_missing_dc = Vec::<(i32, usize, usize)>::new();
+            for (dc, endpoints) in &desired_by_dc {
+                if endpoints.is_empty() {
+                    continue;
+                }
+                let required = Self::required_writers_for_dc(endpoints.len());
+                let fresh_count = writers
+                    .iter()
+                    .filter(|w| !w.draining.load(Ordering::Relaxed))
+                    .filter(|w| w.generation == generation)
+                    .filter(|w| endpoints.contains(&w.addr))
+                    .count();
+                if fresh_count < required {
+                    fresh_missing_dc.push((*dc, fresh_count, required));
+                }
+            }
+            if !fresh_missing_dc.is_empty() {
+                warn!(
+                    previous_generation,
+                    generation,
+                    missing_dc = ?fresh_missing_dc,
+                    "ME hardswap pending: fresh generation coverage incomplete"
+                );
+                return;
+            }
+        } else if !missing_dc.is_empty() {
+            warn!(
+                missing_dc = ?missing_dc,
+                // Keep stale writers alive when fresh coverage is incomplete.
+                "ME reinit coverage incomplete; keeping stale writers"
+            );
+            return;
+        }
+
+        let desired_addrs: HashSet<SocketAddr> = desired_by_dc
+            .values()
+            .flat_map(|set| set.iter().copied())
+            .collect();
+
+        let stale_writer_ids: Vec<u64> = writers
+            .iter()
+            .filter(|w| !w.draining.load(Ordering::Relaxed))
+            .filter(|w| {
+                if hardswap {
+                    w.generation < generation
+                } else {
+                    !desired_addrs.contains(&w.addr)
+                }
+            })
+            .map(|w| w.id)
+            .collect();
+        drop(writers);
+
+        if stale_writer_ids.is_empty() {
+            debug!("ME reinit cycle completed with no stale writers");
+            return;
+        }
+
+        let drain_timeout = self.force_close_timeout();
+        let drain_timeout_secs = drain_timeout.map(|d| d.as_secs()).unwrap_or(0);
+        info!(
+            stale_writers = stale_writer_ids.len(),
+            previous_generation,
+            generation,
+            hardswap,
+            coverage_ratio = format_args!("{coverage_ratio:.3}"),
+            min_ratio = format_args!("{min_ratio:.3}"),
+            drain_timeout_secs,
+            "ME reinit cycle covered; draining stale writers"
+        );
+        self.stats.increment_pool_swap_total();
+        for writer_id in stale_writer_ids {
+            self.mark_writer_draining_with_timeout(writer_id, drain_timeout, !hardswap)
+                .await;
+        }
+    }
+
+    pub async fn zero_downtime_reinit_periodic(self: &Arc<Self>, rng: &SecureRandom) {
+        self.zero_downtime_reinit_after_map_change(rng).await;
+    }
+}

src/transport/middle_proxy/pool_writer.rs

@@ -0,0 +1,366 @@
+use std::net::SocketAddr;
+use std::sync::Arc;
+use std::sync::atomic::{AtomicBool, AtomicU64, Ordering};
+use std::time::{Duration, Instant};
+
+use bytes::BytesMut;
+use rand::Rng;
+use tokio::sync::mpsc;
+use tokio_util::sync::CancellationToken;
+use tracing::{debug, info, warn};
+
+use crate::crypto::SecureRandom;
+use crate::error::{ProxyError, Result};
+use crate::protocol::constants::RPC_PING_U32;
+
+use super::codec::{RpcWriter, WriterCommand};
+use super::pool::{MePool, MeWriter};
+use super::reader::reader_loop;
+use super::registry::BoundConn;
+
+const ME_ACTIVE_PING_SECS: u64 = 25;
+const ME_ACTIVE_PING_JITTER_SECS: i64 = 5;
+const ME_IDLE_KEEPALIVE_MAX_SECS: u64 = 5;
+
+impl MePool {
+    pub(crate) async fn prune_closed_writers(self: &Arc<Self>) {
+        let closed_writer_ids: Vec<u64> = {
+            let ws = self.writers.read().await;
+            ws.iter().filter(|w| w.tx.is_closed()).map(|w| w.id).collect()
+        };
+        if closed_writer_ids.is_empty() {
+            return;
+        }
+
+        for writer_id in closed_writer_ids {
+            if self.registry.is_writer_empty(writer_id).await {
+                let _ = self.remove_writer_only(writer_id).await;
+            } else {
+                let _ = self.remove_writer_and_close_clients(writer_id).await;
+            }
+        }
+    }
+
+    pub(crate) async fn connect_one(self: &Arc<Self>, addr: SocketAddr, rng: &SecureRandom) -> Result<()> {
+        let secret_len = self.proxy_secret.read().await.len();
+        if secret_len < 32 {
+            return Err(ProxyError::Proxy("proxy-secret too short for ME auth".into()));
+        }
+
+        let (stream, _connect_ms, upstream_egress) = self.connect_tcp(addr).await?;
+        let hs = self.handshake_only(stream, addr, upstream_egress, rng).await?;
+
+        let writer_id = self.next_writer_id.fetch_add(1, Ordering::Relaxed);
+        let generation = self.current_generation();
+        let cancel = CancellationToken::new();
+        let degraded = Arc::new(AtomicBool::new(false));
+        let draining = Arc::new(AtomicBool::new(false));
+        let draining_started_at_epoch_secs = Arc::new(AtomicU64::new(0));
+        let allow_drain_fallback = Arc::new(AtomicBool::new(false));
+        let (tx, mut rx) = mpsc::channel::<WriterCommand>(4096);
+        let mut rpc_writer = RpcWriter {
+            writer: hs.wr,
+            key: hs.write_key,
+            iv: hs.write_iv,
+            seq_no: 0,
+            crc_mode: hs.crc_mode,
+        };
+        let cancel_wr = cancel.clone();
+        tokio::spawn(async move {
+            loop {
+                tokio::select! {
+                    cmd = rx.recv() => {
+                        match cmd {
+                            Some(WriterCommand::Data(payload)) => {
+                                if rpc_writer.send(&payload).await.is_err() { break; }
+                            }
+                            Some(WriterCommand::DataAndFlush(payload)) => {
+                                if rpc_writer.send_and_flush(&payload).await.is_err() { break; }
+                            }
+                            Some(WriterCommand::Close) | None => break,
+                        }
+                    }
+                    _ = cancel_wr.cancelled() => break,
+                }
+            }
+        });
+        let writer = MeWriter {
+            id: writer_id,
+            addr,
+            generation,
+            tx: tx.clone(),
+            cancel: cancel.clone(),
+            degraded: degraded.clone(),
+            draining: draining.clone(),
+            draining_started_at_epoch_secs: draining_started_at_epoch_secs.clone(),
+            allow_drain_fallback: allow_drain_fallback.clone(),
+        };
+        self.writers.write().await.push(writer.clone());
+        self.conn_count.fetch_add(1, Ordering::Relaxed);
+        self.writer_available.notify_one();
+
+        let reg = self.registry.clone();
+        let writers_arc = self.writers_arc();
+        let ping_tracker = self.ping_tracker.clone();
+        let ping_tracker_reader = ping_tracker.clone();
+        let rtt_stats = self.rtt_stats.clone();
+        let stats_reader = self.stats.clone();
+        let stats_ping = self.stats.clone();
+        let pool = Arc::downgrade(self);
+        let cancel_ping = cancel.clone();
+        let tx_ping = tx.clone();
+        let ping_tracker_ping = ping_tracker.clone();
+        let cleanup_done = Arc::new(AtomicBool::new(false));
+        let cleanup_for_reader = cleanup_done.clone();
+        let cleanup_for_ping = cleanup_done.clone();
+        let keepalive_enabled = self.me_keepalive_enabled;
+        let keepalive_interval = self.me_keepalive_interval;
+        let keepalive_jitter = self.me_keepalive_jitter;
+        let cancel_reader_token = cancel.clone();
+        let cancel_ping_token = cancel_ping.clone();
+
+        tokio::spawn(async move {
+            let res = reader_loop(
+                hs.rd,
+                hs.read_key,
+                hs.read_iv,
+                hs.crc_mode,
+                reg.clone(),
+                BytesMut::new(),
+                BytesMut::new(),
+                tx.clone(),
+                ping_tracker_reader,
+                rtt_stats.clone(),
+                stats_reader,
+                writer_id,
+                degraded.clone(),
+                cancel_reader_token.clone(),
+            )
+            .await;
+            if let Some(pool) = pool.upgrade()
+                && cleanup_for_reader
+                    .compare_exchange(false, true, Ordering::AcqRel, Ordering::Relaxed)
+                    .is_ok()
+            {
+                pool.remove_writer_and_close_clients(writer_id).await;
+            }
+            if let Err(e) = res {
+                warn!(error = %e, "ME reader ended");
+            }
+            let mut ws = writers_arc.write().await;
+            ws.retain(|w| w.id != writer_id);
+            info!(remaining = ws.len(), "Dead ME writer removed from pool");
+        });
+
+        let pool_ping = Arc::downgrade(self);
+        tokio::spawn(async move {
+            let mut ping_id: i64 = rand::random::<i64>();
+            let idle_interval_cap = Duration::from_secs(ME_IDLE_KEEPALIVE_MAX_SECS);
+            // Per-writer jittered start to avoid phase sync.
+            let startup_jitter = if keepalive_enabled {
+                let mut interval = keepalive_interval;
+                if let Some(pool) = pool_ping.upgrade() {
+                    if pool.registry.is_writer_empty(writer_id).await {
+                        interval = interval.min(idle_interval_cap);
+                    }
+                } else {
+                    return;
+                }
+                let jitter_cap_ms = interval.as_millis() / 2;
+                let effective_jitter_ms = keepalive_jitter.as_millis().min(jitter_cap_ms).max(1);
+                Duration::from_millis(rand::rng().random_range(0..=effective_jitter_ms as u64))
+            } else {
+                let jitter = rand::rng().random_range(-ME_ACTIVE_PING_JITTER_SECS..=ME_ACTIVE_PING_JITTER_SECS);
+                let wait = (ME_ACTIVE_PING_SECS as i64 + jitter).max(5) as u64;
+                Duration::from_secs(wait)
+            };
+            tokio::select! {
+                _ = cancel_ping_token.cancelled() => return,
+                _ = tokio::time::sleep(startup_jitter) => {}
+            }
+            loop {
+                let wait = if keepalive_enabled {
+                    let mut interval = keepalive_interval;
+                    if let Some(pool) = pool_ping.upgrade() {
+                        if pool.registry.is_writer_empty(writer_id).await {
+                            interval = interval.min(idle_interval_cap);
+                        }
+                    } else {
+                        break;
+                    }
+                    let jitter_cap_ms = interval.as_millis() / 2;
+                    let effective_jitter_ms = keepalive_jitter.as_millis().min(jitter_cap_ms).max(1);
+                    interval + Duration::from_millis(rand::rng().random_range(0..=effective_jitter_ms as u64))
+                } else {
+                    let jitter = rand::rng().random_range(-ME_ACTIVE_PING_JITTER_SECS..=ME_ACTIVE_PING_JITTER_SECS);
+                    let secs = (ME_ACTIVE_PING_SECS as i64 + jitter).max(5) as u64;
+                    Duration::from_secs(secs)
+                };
+                tokio::select! {
+                    _ = cancel_ping_token.cancelled() => {
+                        break;
+                    }
+                    _ = tokio::time::sleep(wait) => {}
+                }
+                let sent_id = ping_id;
+                let mut p = Vec::with_capacity(12);
+                p.extend_from_slice(&RPC_PING_U32.to_le_bytes());
+                p.extend_from_slice(&sent_id.to_le_bytes());
+                {
+                    let mut tracker = ping_tracker_ping.lock().await;
+                    let before = tracker.len();
+                    tracker.retain(|_, (ts, _)| ts.elapsed() < Duration::from_secs(120));
+                    let expired = before.saturating_sub(tracker.len());
+                    if expired > 0 {
+                        stats_ping.increment_me_keepalive_timeout_by(expired as u64);
+                    }
+                    tracker.insert(sent_id, (std::time::Instant::now(), writer_id));
+                }
+                ping_id = ping_id.wrapping_add(1);
+                stats_ping.increment_me_keepalive_sent();
+                if tx_ping.send(WriterCommand::DataAndFlush(p)).await.is_err() {
+                    stats_ping.increment_me_keepalive_failed();
+                    debug!("ME ping failed, removing dead writer");
+                    cancel_ping.cancel();
+                    if let Some(pool) = pool_ping.upgrade()
+                        && cleanup_for_ping
+                            .compare_exchange(false, true, Ordering::AcqRel, Ordering::Relaxed)
+                            .is_ok()
+                    {
+                        pool.remove_writer_and_close_clients(writer_id).await;
+                    }
+                    break;
+                }
+            }
+        });
+
+        Ok(())
+    }
+
+    pub(crate) async fn remove_writer_and_close_clients(self: &Arc<Self>, writer_id: u64) {
+        let conns = self.remove_writer_only(writer_id).await;
+        for bound in conns {
+            let _ = self.registry.route(bound.conn_id, super::MeResponse::Close).await;
+            let _ = self.registry.unregister(bound.conn_id).await;
+        }
+    }
+
+    async fn remove_writer_only(self: &Arc<Self>, writer_id: u64) -> Vec<BoundConn> {
+        let mut close_tx: Option<mpsc::Sender<WriterCommand>> = None;
+        let mut removed_addr: Option<SocketAddr> = None;
+        let mut trigger_refill = false;
+        {
+            let mut ws = self.writers.write().await;
+            if let Some(pos) = ws.iter().position(|w| w.id == writer_id) {
+                let w = ws.remove(pos);
+                let was_draining = w.draining.load(Ordering::Relaxed);
+                if was_draining {
+                    self.stats.decrement_pool_drain_active();
+                }
+                self.stats.increment_me_writer_removed_total();
+                w.cancel.cancel();
+                removed_addr = Some(w.addr);
+                trigger_refill = !was_draining;
+                if trigger_refill {
+                    self.stats.increment_me_writer_removed_unexpected_total();
+                }
+                close_tx = Some(w.tx.clone());
+                self.conn_count.fetch_sub(1, Ordering::Relaxed);
+            }
+        }
+        if let Some(tx) = close_tx {
+            let _ = tx.send(WriterCommand::Close).await;
+        }
+        if trigger_refill
+            && let Some(addr) = removed_addr
+        {
+            self.trigger_immediate_refill(addr);
+        }
+        self.rtt_stats.lock().await.remove(&writer_id);
+        self.registry.writer_lost(writer_id).await
+    }
+
+    pub(crate) async fn mark_writer_draining_with_timeout(
+        self: &Arc<Self>,
+        writer_id: u64,
+        timeout: Option<Duration>,
+        allow_drain_fallback: bool,
+    ) {
+        let timeout = timeout.filter(|d| !d.is_zero());
+        let found = {
+            let mut ws = self.writers.write().await;
+            if let Some(w) = ws.iter_mut().find(|w| w.id == writer_id) {
+                let already_draining = w.draining.swap(true, Ordering::Relaxed);
+                w.allow_drain_fallback
+                    .store(allow_drain_fallback, Ordering::Relaxed);
+                w.draining_started_at_epoch_secs
+                    .store(Self::now_epoch_secs(), Ordering::Relaxed);
+                if !already_draining {
+                    self.stats.increment_pool_drain_active();
+                }
+                w.draining.store(true, Ordering::Relaxed);
+                true
+            } else {
+                false
+            }
+        };
+
+        if !found {
+            return;
+        }
+
+        let timeout_secs = timeout.map(|d| d.as_secs()).unwrap_or(0);
+        debug!(
+            writer_id,
+            timeout_secs,
+            allow_drain_fallback,
+            "ME writer marked draining"
+        );
+
+        let pool = Arc::downgrade(self);
+        tokio::spawn(async move {
+            let deadline = timeout.map(|t| Instant::now() + t);
+            while let Some(p) = pool.upgrade() {
+                if let Some(deadline_at) = deadline
+                    && Instant::now() >= deadline_at
+                {
+                    warn!(writer_id, "Drain timeout, force-closing");
+                    p.stats.increment_pool_force_close_total();
+                    let _ = p.remove_writer_and_close_clients(writer_id).await;
+                    break;
+                }
+                if p.registry.is_writer_empty(writer_id).await {
+                    let _ = p.remove_writer_only(writer_id).await;
+                    break;
+                }
+                tokio::time::sleep(Duration::from_secs(1)).await;
+            }
+        });
+    }
+
+    pub(crate) async fn mark_writer_draining(self: &Arc<Self>, writer_id: u64) {
+        self.mark_writer_draining_with_timeout(writer_id, Some(Duration::from_secs(300)), false)
+            .await;
+    }
+
+    pub(super) fn writer_accepts_new_binding(&self, writer: &MeWriter) -> bool {
+        if !writer.draining.load(Ordering::Relaxed) {
+            return true;
+        }
+        if !writer.allow_drain_fallback.load(Ordering::Relaxed) {
+            return false;
+        }
+
+        let ttl_secs = self.me_pool_drain_ttl_secs.load(Ordering::Relaxed);
+        if ttl_secs == 0 {
+            return true;
+        }
+
+        let started = writer.draining_started_at_epoch_secs.load(Ordering::Relaxed);
+        if started == 0 {
+            return false;
+        }
+
+        Self::now_epoch_secs().saturating_sub(started) <= ttl_secs
+    }
+}

src/transport/middle_proxy/reader.rs

@@ -124,7 +124,14 @@ pub(crate) async fn reader_loop(
                     match routed {
                         RouteResult::NoConn => stats.increment_me_route_drop_no_conn(),
                         RouteResult::ChannelClosed => stats.increment_me_route_drop_channel_closed(),
-                        RouteResult::QueueFull => stats.increment_me_route_drop_queue_full(),
+                        RouteResult::QueueFullBase => {
+                            stats.increment_me_route_drop_queue_full();
+                            stats.increment_me_route_drop_queue_full_base();
+                        }
+                        RouteResult::QueueFullHigh => {
+                            stats.increment_me_route_drop_queue_full();
+                            stats.increment_me_route_drop_queue_full_high();
+                        }
                         RouteResult::Routed => {}
                     }
                     reg.unregister(cid).await;
@@ -140,7 +147,14 @@ pub(crate) async fn reader_loop(
                     match routed {
                         RouteResult::NoConn => stats.increment_me_route_drop_no_conn(),
                         RouteResult::ChannelClosed => stats.increment_me_route_drop_channel_closed(),
-                        RouteResult::QueueFull => stats.increment_me_route_drop_queue_full(),
+                        RouteResult::QueueFullBase => {
+                            stats.increment_me_route_drop_queue_full();
+                            stats.increment_me_route_drop_queue_full_base();
+                        }
+                        RouteResult::QueueFullHigh => {
+                            stats.increment_me_route_drop_queue_full();
+                            stats.increment_me_route_drop_queue_full_high();
+                        }
                         RouteResult::Routed => {}
                     }
                     reg.unregister(cid).await;

src/transport/middle_proxy/registry.rs

@@ -1,6 +1,6 @@
 use std::collections::{HashMap, HashSet};
 use std::net::SocketAddr;
-use std::sync::atomic::{AtomicU64, Ordering};
+use std::sync::atomic::{AtomicU8, AtomicU64, Ordering};
 use std::time::Duration;
 
 use tokio::sync::{mpsc, RwLock};
@@ -10,14 +10,17 @@ use super::codec::WriterCommand;
 use super::MeResponse;
 
 const ROUTE_CHANNEL_CAPACITY: usize = 4096;
-const ROUTE_BACKPRESSURE_TIMEOUT: Duration = Duration::from_millis(25);
+const ROUTE_BACKPRESSURE_BASE_TIMEOUT_MS: u64 = 25;
+const ROUTE_BACKPRESSURE_HIGH_TIMEOUT_MS: u64 = 120;
+const ROUTE_BACKPRESSURE_HIGH_WATERMARK_PCT: u8 = 80;
 
 #[derive(Debug, Clone, Copy, PartialEq, Eq)]
 pub enum RouteResult {
     Routed,
     NoConn,
     ChannelClosed,
-    QueueFull,
+    QueueFullBase,
+    QueueFullHigh,
 }
 
 #[derive(Clone)]
@@ -65,6 +68,9 @@ impl RegistryInner {
 pub struct ConnRegistry {
     inner: RwLock<RegistryInner>,
     next_id: AtomicU64,
+    route_backpressure_base_timeout_ms: AtomicU64,
+    route_backpressure_high_timeout_ms: AtomicU64,
+    route_backpressure_high_watermark_pct: AtomicU8,
 }
 
 impl ConnRegistry {
@@ -73,9 +79,35 @@ impl ConnRegistry {
         Self {
             inner: RwLock::new(RegistryInner::new()),
             next_id: AtomicU64::new(start),
+            route_backpressure_base_timeout_ms: AtomicU64::new(
+                ROUTE_BACKPRESSURE_BASE_TIMEOUT_MS,
+            ),
+            route_backpressure_high_timeout_ms: AtomicU64::new(
+                ROUTE_BACKPRESSURE_HIGH_TIMEOUT_MS,
+            ),
+            route_backpressure_high_watermark_pct: AtomicU8::new(
+                ROUTE_BACKPRESSURE_HIGH_WATERMARK_PCT,
+            ),
         }
     }
 
+    pub fn update_route_backpressure_policy(
+        &self,
+        base_timeout_ms: u64,
+        high_timeout_ms: u64,
+        high_watermark_pct: u8,
+    ) {
+        let base = base_timeout_ms.max(1);
+        let high = high_timeout_ms.max(base);
+        let watermark = high_watermark_pct.clamp(1, 100);
+        self.route_backpressure_base_timeout_ms
+            .store(base, Ordering::Relaxed);
+        self.route_backpressure_high_timeout_ms
+            .store(high, Ordering::Relaxed);
+        self.route_backpressure_high_watermark_pct
+            .store(watermark, Ordering::Relaxed);
+    }
+
     pub async fn register(&self) -> (u64, mpsc::Receiver<MeResponse>) {
         let id = self.next_id.fetch_add(1, Ordering::Relaxed);
         let (tx, rx) = mpsc::channel(ROUTE_CHANNEL_CAPACITY);
@@ -112,10 +144,40 @@ impl ConnRegistry {
             Err(TrySendError::Closed(_)) => RouteResult::ChannelClosed,
             Err(TrySendError::Full(resp)) => {
                 // Absorb short bursts without dropping/closing the session immediately.
-                match tokio::time::timeout(ROUTE_BACKPRESSURE_TIMEOUT, tx.send(resp)).await {
+                let base_timeout_ms =
+                    self.route_backpressure_base_timeout_ms.load(Ordering::Relaxed).max(1);
+                let high_timeout_ms = self
+                    .route_backpressure_high_timeout_ms
+                    .load(Ordering::Relaxed)
+                    .max(base_timeout_ms);
+                let high_watermark_pct = self
+                    .route_backpressure_high_watermark_pct
+                    .load(Ordering::Relaxed)
+                    .clamp(1, 100);
+                let used = ROUTE_CHANNEL_CAPACITY.saturating_sub(tx.capacity());
+                let used_pct = if ROUTE_CHANNEL_CAPACITY == 0 {
+                    100
+                } else {
+                    (used.saturating_mul(100) / ROUTE_CHANNEL_CAPACITY) as u8
+                };
+                let high_profile = used_pct >= high_watermark_pct;
+                let timeout_ms = if high_profile {
+                    high_timeout_ms
+                } else {
+                    base_timeout_ms
+                };
+                let timeout_dur = Duration::from_millis(timeout_ms);
+
+                match tokio::time::timeout(timeout_dur, tx.send(resp)).await {
                     Ok(Ok(())) => RouteResult::Routed,
                     Ok(Err(_)) => RouteResult::ChannelClosed,
-                    Err(_) => RouteResult::QueueFull,
+                    Err(_) => {
+                        if high_profile {
+                            RouteResult::QueueFullHigh
+                        } else {
+                            RouteResult::QueueFullBase
+                        }
+                    }
                 }
             }
         }

src/transport/middle_proxy/rotation.rs

@@ -1,50 +1,87 @@
 use std::sync::Arc;
-use std::sync::atomic::Ordering;
 use std::time::Duration;
 
+use tokio::sync::watch;
 use tracing::{info, warn};
 
+use crate::config::ProxyConfig;
 use crate::crypto::SecureRandom;
 
 use super::MePool;
 
-/// Periodically refresh ME connections to avoid long-lived degradation.
-pub async fn me_rotation_task(pool: Arc<MePool>, rng: Arc<SecureRandom>, interval: Duration) {
-    let interval = interval.max(Duration::from_secs(600));
+/// Periodically reinitialize ME generations and swap them after full warmup.
+pub async fn me_rotation_task(
+    pool: Arc<MePool>,
+    rng: Arc<SecureRandom>,
+    mut config_rx: watch::Receiver<Arc<ProxyConfig>>,
+) {
+    let mut interval_secs = config_rx
+        .borrow()
+        .general
+        .effective_me_reinit_every_secs()
+        .max(1);
+    let mut interval = Duration::from_secs(interval_secs);
+    let mut next_tick = tokio::time::Instant::now() + interval;
+
+    info!(interval_secs, "ME periodic reinit task started");
+
     loop {
-        tokio::time::sleep(interval).await;
+        let sleep = tokio::time::sleep_until(next_tick);
+        tokio::pin!(sleep);
 
-        let candidate = {
-            let ws = pool.writers.read().await;
-            if ws.is_empty() {
-                None
-            } else {
-                let idx = (pool.rr.load(std::sync::atomic::Ordering::Relaxed) as usize) % ws.len();
-                ws.get(idx).cloned()
-            }
-        };
-
-        let Some(w) = candidate else {
-            continue;
-        };
-
-        info!(addr = %w.addr, writer_id = w.id, "Rotating ME connection");
-        match pool.connect_one(w.addr, rng.as_ref()).await {
-            Ok(()) => {
-                tokio::time::sleep(Duration::from_secs(2)).await;
-                let ws = pool.writers.read().await;
-                let new_alive = ws.iter().any(|nw|
-                    nw.id != w.id && nw.addr == w.addr && !nw.degraded.load(Ordering::Relaxed) && !nw.draining.load(Ordering::Relaxed)
-                );
-                drop(ws);
-                if new_alive {
-                    pool.mark_writer_draining(w.id).await;
-                } else {
-                    warn!(addr = %w.addr, writer_id = w.id, "New writer died, keeping old");
+        tokio::select! {
+            _ = &mut sleep => {
+                pool.zero_downtime_reinit_periodic(rng.as_ref()).await;
+                let refreshed_secs = config_rx
+                    .borrow()
+                    .general
+                    .effective_me_reinit_every_secs()
+                    .max(1);
+                if refreshed_secs != interval_secs {
+                    info!(
+                        old_me_reinit_every_secs = interval_secs,
+                        new_me_reinit_every_secs = refreshed_secs,
+                        "ME periodic reinit interval changed"
+                    );
+                    interval_secs = refreshed_secs;
+                    interval = Duration::from_secs(interval_secs);
                 }
+                next_tick = tokio::time::Instant::now() + interval;
             }
-            Err(e) => {
-                warn!(addr = %w.addr, writer_id = w.id, error = %e, "ME rotation connect failed");
+            changed = config_rx.changed() => {
+                if changed.is_err() {
+                    warn!("ME periodic reinit task stopped: config channel closed");
+                    break;
+                }
+                let new_secs = config_rx
+                    .borrow()
+                    .general
+                    .effective_me_reinit_every_secs()
+                    .max(1);
+                if new_secs == interval_secs {
+                    continue;
+                }
+
+                if new_secs < interval_secs {
+                    info!(
+                        old_me_reinit_every_secs = interval_secs,
+                        new_me_reinit_every_secs = new_secs,
+                        "ME periodic reinit interval decreased, running immediate reinit"
+                    );
+                    interval_secs = new_secs;
+                    interval = Duration::from_secs(interval_secs);
+                    pool.zero_downtime_reinit_periodic(rng.as_ref()).await;
+                    next_tick = tokio::time::Instant::now() + interval;
+                } else {
+                    info!(
+                        old_me_reinit_every_secs = interval_secs,
+                        new_me_reinit_every_secs = new_secs,
+                        "ME periodic reinit interval increased"
+                    );
+                    interval_secs = new_secs;
+                    interval = Duration::from_secs(interval_secs);
+                    next_tick = tokio::time::Instant::now() + interval;
+                }
             }
         }
     }

src/transport/mod.rs

@@ -14,5 +14,5 @@ pub use socket::*;
 #[allow(unused_imports)]
 pub use socks::*;
 #[allow(unused_imports)]
-pub use upstream::{DcPingResult, StartupPingResult, UpstreamManager};
+pub use upstream::{DcPingResult, StartupPingResult, UpstreamEgressInfo, UpstreamManager, UpstreamRouteKind};
 pub mod middle_proxy;

src/transport/proxy_protocol.rs

@@ -233,14 +233,12 @@ async fn parse_v2<R: AsyncRead + Unpin>(
 }
 
 /// Builder for PROXY protocol v1 header
-#[allow(dead_code)]
 pub struct ProxyProtocolV1Builder {
     family: &'static str,
     src_addr: Option<SocketAddr>,
     dst_addr: Option<SocketAddr>,
 }
 
-#[allow(dead_code)]
 impl ProxyProtocolV1Builder {
     pub fn new() -> Self {
         Self {
@@ -288,13 +286,17 @@ impl Default for ProxyProtocolV1Builder {
 }
 
 /// Builder for PROXY protocol v2 header
-#[allow(dead_code)]
 pub struct ProxyProtocolV2Builder {
     src: Option<SocketAddr>,
     dst: Option<SocketAddr>,
 }
 
-#[allow(dead_code)]
+impl Default for ProxyProtocolV2Builder {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
 impl ProxyProtocolV2Builder {
     pub fn new() -> Self {
         Self { src: None, dst: None }

src/transport/socks.rs

@@ -5,11 +5,16 @@ use tokio::io::{AsyncReadExt, AsyncWriteExt};
 use tokio::net::TcpStream;
 use crate::error::{ProxyError, Result};
 
+#[derive(Debug, Clone, Copy)]
+pub struct SocksBoundAddr {
+    pub addr: SocketAddr,
+}
+
 pub async fn connect_socks4(
     stream: &mut TcpStream,
     target: SocketAddr,
     user_id: Option<&str>,
-) -> Result<()> {
+) -> Result<SocksBoundAddr> {
     let ip = match target.ip() {
         IpAddr::V4(ip) => ip,
         IpAddr::V6(_) => return Err(ProxyError::Proxy("SOCKS4 does not support IPv6".to_string())),
@@ -36,8 +41,13 @@ pub async fn connect_socks4(
     if resp[1] != 90 {
         return Err(ProxyError::Proxy(format!("SOCKS4 request rejected: code {}", resp[1])));
     }
-    
-    Ok(())
+
+    let bound_port = u16::from_be_bytes([resp[2], resp[3]]);
+    let bound_ip = IpAddr::from([resp[4], resp[5], resp[6], resp[7]]);
+
+    Ok(SocksBoundAddr {
+        addr: SocketAddr::new(bound_ip, bound_port),
+    })
 }
 
 pub async fn connect_socks5(
@@ -45,7 +55,7 @@ pub async fn connect_socks5(
     target: SocketAddr,
     username: Option<&str>,
     password: Option<&str>,
-) -> Result<()> {
+) -> Result<SocksBoundAddr> {
     // 1. Auth negotiation
     // VER (1) | NMETHODS (1) | METHODS (variable)
     let mut methods = vec![0u8]; // No auth
@@ -122,24 +132,36 @@ pub async fn connect_socks5(
         return Err(ProxyError::Proxy(format!("SOCKS5 request failed: code {}", head[1])));
     }
     
-    // Skip address part of response
-    match head[3] {
+    // Parse bound address from response.
+    let bound_addr = match head[3] {
         1 => { // IPv4
             let mut addr = [0u8; 4 + 2];
             stream.read_exact(&mut addr).await.map_err(ProxyError::Io)?;
+            let ip = IpAddr::from([addr[0], addr[1], addr[2], addr[3]]);
+            let port = u16::from_be_bytes([addr[4], addr[5]]);
+            SocketAddr::new(ip, port)
         },
         3 => { // Domain
             let mut len = [0u8; 1];
             stream.read_exact(&mut len).await.map_err(ProxyError::Io)?;
             let mut addr = vec![0u8; len[0] as usize + 2];
             stream.read_exact(&mut addr).await.map_err(ProxyError::Io)?;
+            // Domain-bound response is not useful for KDF IP material.
+            let port_pos = addr.len().saturating_sub(2);
+            let port = u16::from_be_bytes([addr[port_pos], addr[port_pos + 1]]);
+            SocketAddr::new(IpAddr::from([0, 0, 0, 0]), port)
         },
         4 => { // IPv6
             let mut addr = [0u8; 16 + 2];
             stream.read_exact(&mut addr).await.map_err(ProxyError::Io)?;
+            let ip = IpAddr::from(<[u8; 16]>::try_from(&addr[..16]).map_err(|_| {
+                ProxyError::Proxy("Invalid SOCKS5 IPv6 bound address".to_string())
+            })?);
+            let port = u16::from_be_bytes([addr[16], addr[17]]);
+            SocketAddr::new(ip, port)
         },
         _ => return Err(ProxyError::Proxy("Invalid address type in SOCKS5 response".to_string())),
-    }
-    
-    Ok(())
-}
+    };
+
+    Ok(SocksBoundAddr { addr: bound_addr })
+}

src/transport/upstream.rs

@@ -17,6 +17,7 @@ use tracing::{debug, warn, info, trace};
 
 use crate::config::{UpstreamConfig, UpstreamType};
 use crate::error::{Result, ProxyError};
+use crate::network::dns_overrides::{resolve_socket_addr, split_host_port};
 use crate::protocol::constants::{TG_DATACENTERS_V4, TG_DATACENTERS_V6, TG_DATACENTER_PORT};
 use crate::transport::socket::{create_outgoing_socket_bound, resolve_interface_ip};
 use crate::transport::socks::{connect_socks4, connect_socks5};
@@ -150,15 +151,39 @@ pub struct StartupPingResult {
     pub both_available: bool,
 }
 
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub enum UpstreamRouteKind {
+    Direct,
+    Socks4,
+    Socks5,
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub struct UpstreamEgressInfo {
+    pub route_kind: UpstreamRouteKind,
+    pub local_addr: Option<SocketAddr>,
+    pub direct_bind_ip: Option<IpAddr>,
+    pub socks_bound_addr: Option<SocketAddr>,
+    pub socks_proxy_addr: Option<SocketAddr>,
+}
+
 // ============= Upstream Manager =============
 
 #[derive(Clone)]
 pub struct UpstreamManager {
     upstreams: Arc<RwLock<Vec<UpstreamState>>>,
+    connect_retry_attempts: u32,
+    connect_retry_backoff: Duration,
+    unhealthy_fail_threshold: u32,
 }
 
 impl UpstreamManager {
-    pub fn new(configs: Vec<UpstreamConfig>) -> Self {
+    pub fn new(
+        configs: Vec<UpstreamConfig>,
+        connect_retry_attempts: u32,
+        connect_retry_backoff_ms: u64,
+        unhealthy_fail_threshold: u32,
+    ) -> Self {
         let states = configs.into_iter()
             .filter(|c| c.enabled)
             .map(UpstreamState::new)
@@ -166,24 +191,88 @@ impl UpstreamManager {
 
         Self {
             upstreams: Arc::new(RwLock::new(states)),
+            connect_retry_attempts: connect_retry_attempts.max(1),
+            connect_retry_backoff: Duration::from_millis(connect_retry_backoff_ms),
+            unhealthy_fail_threshold: unhealthy_fail_threshold.max(1),
         }
     }
 
+    #[cfg(unix)]
+    fn resolve_interface_addrs(name: &str, want_ipv6: bool) -> Vec<IpAddr> {
+        use nix::ifaddrs::getifaddrs;
+
+        let mut out = Vec::new();
+        if let Ok(addrs) = getifaddrs() {
+            for iface in addrs {
+                if iface.interface_name != name {
+                    continue;
+                }
+                if let Some(address) = iface.address {
+                    if let Some(v4) = address.as_sockaddr_in() {
+                        if !want_ipv6 {
+                            out.push(IpAddr::V4(v4.ip()));
+                        }
+                    } else if let Some(v6) = address.as_sockaddr_in6()
+                        && want_ipv6
+                    {
+                        out.push(IpAddr::V6(v6.ip()));
+                    }
+                }
+            }
+        }
+        out.sort_unstable();
+        out.dedup();
+        out
+    }
+
     fn resolve_bind_address(
         interface: &Option<String>,
         bind_addresses: &Option<Vec<String>>,
         target: SocketAddr,
         rr: Option<&AtomicUsize>,
+        validate_ip_on_interface: bool,
     ) -> Option<IpAddr> {
         let want_ipv6 = target.is_ipv6();
 
         if let Some(addrs) = bind_addresses {
-            let candidates: Vec<IpAddr> = addrs
+            let mut candidates: Vec<IpAddr> = addrs
                 .iter()
                 .filter_map(|s| s.parse::<IpAddr>().ok())
                 .filter(|ip| ip.is_ipv6() == want_ipv6)
                 .collect();
 
+            // Explicit bind IP has strict priority over interface auto-selection.
+            if validate_ip_on_interface
+                && let Some(iface) = interface
+                && iface.parse::<IpAddr>().is_err()
+            {
+                #[cfg(unix)]
+                {
+                    let iface_addrs = Self::resolve_interface_addrs(iface, want_ipv6);
+                    if !iface_addrs.is_empty() {
+                        candidates.retain(|ip| {
+                            let ok = iface_addrs.contains(ip);
+                            if !ok {
+                                warn!(
+                                    interface = %iface,
+                                    bind_ip = %ip,
+                                    target = %target,
+                                    "Configured bind address is not assigned to interface"
+                                );
+                            }
+                            ok
+                        });
+                    } else if !candidates.is_empty() {
+                        warn!(
+                            interface = %iface,
+                            target = %target,
+                            "Configured interface has no addresses for target family; falling back to direct connect without bind"
+                        );
+                        candidates.clear();
+                    }
+                }
+            }
+
             if !candidates.is_empty() {
                 if let Some(counter) = rr {
                     let idx = counter.fetch_add(1, Ordering::Relaxed) % candidates.len();
@@ -191,6 +280,19 @@ impl UpstreamManager {
                 }
                 return candidates.first().copied();
             }
+
+            if validate_ip_on_interface
+                && interface
+                    .as_ref()
+                    .is_some_and(|iface| iface.parse::<IpAddr>().is_err())
+            {
+                warn!(
+                    interface = interface.as_deref().unwrap_or(""),
+                    target = %target,
+                    "No valid bind_addresses left for interface; falling back to direct connect without bind"
+                );
+                return None;
+            }
         }
 
         if let Some(iface) = interface {
@@ -209,6 +311,31 @@ impl UpstreamManager {
         None
     }
 
+    async fn connect_hostname_with_dns_override(
+        address: &str,
+        connect_timeout: Duration,
+    ) -> Result<TcpStream> {
+        if let Some((host, port)) = split_host_port(address)
+            && let Some(addr) = resolve_socket_addr(&host, port)
+        {
+            return match tokio::time::timeout(connect_timeout, TcpStream::connect(addr)).await {
+                Ok(Ok(stream)) => Ok(stream),
+                Ok(Err(e)) => Err(ProxyError::Io(e)),
+                Err(_) => Err(ProxyError::ConnectionTimeout {
+                    addr: addr.to_string(),
+                }),
+            };
+        }
+
+        match tokio::time::timeout(connect_timeout, TcpStream::connect(address)).await {
+            Ok(Ok(stream)) => Ok(stream),
+            Ok(Err(e)) => Err(ProxyError::Io(e)),
+            Err(_) => Err(ProxyError::ConnectionTimeout {
+                addr: address.to_string(),
+            }),
+        }
+    }
+
     /// Select upstream using latency-weighted random selection.
     async fn select_upstream(&self, dc_idx: Option<i16>, scope: Option<&str>) -> Option<usize> {
         let upstreams = self.upstreams.read().await;
@@ -290,6 +417,17 @@ impl UpstreamManager {
 
     /// Connect to target through a selected upstream.
     pub async fn connect(&self, target: SocketAddr, dc_idx: Option<i16>, scope: Option<&str>) -> Result<TcpStream> {
+        let (stream, _) = self.connect_with_details(target, dc_idx, scope).await?;
+        Ok(stream)
+    }
+
+    /// Connect to target through a selected upstream and return egress details.
+    pub async fn connect_with_details(
+        &self,
+        target: SocketAddr,
+        dc_idx: Option<i16>,
+        scope: Option<&str>,
+    ) -> Result<(TcpStream, UpstreamEgressInfo)> {
         let idx = self.select_upstream(dc_idx, scope).await
             .ok_or_else(|| ProxyError::Config("No upstreams available".to_string()))?;
 
@@ -303,43 +441,83 @@ impl UpstreamManager {
             upstream.selected_scope = s.to_string();
         }
 
-        let start = Instant::now();
-
         let bind_rr = {
             let guard = self.upstreams.read().await;
             guard.get(idx).map(|u| u.bind_rr.clone())
         };
 
-        match self.connect_via_upstream(&upstream, target, bind_rr).await {
-            Ok(stream) => {
-                let rtt_ms = start.elapsed().as_secs_f64() * 1000.0;
-                let mut guard = self.upstreams.write().await;
-                if let Some(u) = guard.get_mut(idx) {
-                    if !u.healthy {
-                        debug!(rtt_ms = format!("{:.1}", rtt_ms), "Upstream recovered");
-                    }
-                    u.healthy = true;
-                    u.fails = 0;
+        let mut last_error: Option<ProxyError> = None;
+        for attempt in 1..=self.connect_retry_attempts {
+            let start = Instant::now();
+            match self
+                .connect_via_upstream(&upstream, target, bind_rr.clone())
+                .await
+            {
+                Ok((stream, egress)) => {
+                    let rtt_ms = start.elapsed().as_secs_f64() * 1000.0;
+                    let mut guard = self.upstreams.write().await;
+                    if let Some(u) = guard.get_mut(idx) {
+                        if !u.healthy {
+                            debug!(rtt_ms = format!("{:.1}", rtt_ms), "Upstream recovered");
+                        }
+                        if attempt > 1 {
+                            debug!(
+                                attempt,
+                                attempts = self.connect_retry_attempts,
+                                rtt_ms = format!("{:.1}", rtt_ms),
+                                "Upstream connect recovered after retry"
+                            );
+                        }
+                        u.healthy = true;
+                        u.fails = 0;
 
-                    if let Some(di) = dc_idx.and_then(UpstreamState::dc_array_idx) {
-                        u.dc_latency[di].update(rtt_ms);
+                        if let Some(di) = dc_idx.and_then(UpstreamState::dc_array_idx) {
+                            u.dc_latency[di].update(rtt_ms);
+                        }
                     }
+                    return Ok((stream, egress));
                 }
-                Ok(stream)
-            },
-            Err(e) => {
-                let mut guard = self.upstreams.write().await;
-                if let Some(u) = guard.get_mut(idx) {
-                    u.fails += 1;
-                    warn!(fails = u.fails, "Upstream failed: {}", e);
-                    if u.fails > 3 {
-                        u.healthy = false;
-                        warn!("Upstream marked unhealthy");
+                Err(e) => {
+                    if attempt < self.connect_retry_attempts {
+                        debug!(
+                            attempt,
+                            attempts = self.connect_retry_attempts,
+                            target = %target,
+                            error = %e,
+                            "Upstream connect attempt failed, retrying"
+                        );
+                        if !self.connect_retry_backoff.is_zero() {
+                            tokio::time::sleep(self.connect_retry_backoff).await;
+                        }
                     }
+                    last_error = Some(e);
                 }
-                Err(e)
             }
         }
+
+        let error = last_error.unwrap_or_else(|| {
+            ProxyError::Config("Upstream connect attempts exhausted".to_string())
+        });
+
+        let mut guard = self.upstreams.write().await;
+        if let Some(u) = guard.get_mut(idx) {
+            u.fails += 1;
+            warn!(
+                fails = u.fails,
+                attempts = self.connect_retry_attempts,
+                "Upstream failed after retries: {}",
+                error
+            );
+            if u.fails >= self.unhealthy_fail_threshold {
+                u.healthy = false;
+                warn!(
+                    fails = u.fails,
+                    threshold = self.unhealthy_fail_threshold,
+                    "Upstream marked unhealthy"
+                );
+            }
+        }
+        Err(error)
     }
 
     async fn connect_via_upstream(
@@ -347,7 +525,7 @@ impl UpstreamManager {
         config: &UpstreamConfig,
         target: SocketAddr,
         bind_rr: Option<Arc<AtomicUsize>>,
-    ) -> Result<TcpStream> {
+    ) -> Result<(TcpStream, UpstreamEgressInfo)> {
         match &config.upstream_type {
             UpstreamType::Direct { interface, bind_addresses } => {
                 let bind_ip = Self::resolve_bind_address(
@@ -355,6 +533,7 @@ impl UpstreamManager {
                     bind_addresses,
                     target,
                     bind_rr.as_deref(),
+                    true,
                 );
 
                 let socket = create_outgoing_socket_bound(target, bind_ip)?;
@@ -388,7 +567,17 @@ impl UpstreamManager {
                     return Err(ProxyError::Io(e));
                 }
 
-                Ok(stream)
+                let local_addr = stream.local_addr().ok();
+                Ok((
+                    stream,
+                    UpstreamEgressInfo {
+                        route_kind: UpstreamRouteKind::Direct,
+                        local_addr,
+                        direct_bind_ip: bind_ip,
+                        socks_bound_addr: None,
+                        socks_proxy_addr: None,
+                    },
+                ))
             },
             UpstreamType::Socks4 { address, interface, user_id } => {
                 let connect_timeout = Duration::from_secs(DIRECT_CONNECT_TIMEOUT_SECS);
@@ -400,6 +589,7 @@ impl UpstreamManager {
                         &None,
                         proxy_addr,
                         bind_rr.as_deref(),
+                        false,
                     );
 
                     let socket = create_outgoing_socket_bound(proxy_addr, bind_ip)?;
@@ -433,15 +623,7 @@ impl UpstreamManager {
                     if interface.is_some() {
                         warn!("SOCKS4 interface binding is not supported for hostname addresses, ignoring");
                     }
-                    match tokio::time::timeout(connect_timeout, TcpStream::connect(address)).await {
-                        Ok(Ok(stream)) => stream,
-                        Ok(Err(e)) => return Err(ProxyError::Io(e)),
-                        Err(_) => {
-                            return Err(ProxyError::ConnectionTimeout {
-                                addr: address.clone(),
-                            });
-                        }
-                    }
+                    Self::connect_hostname_with_dns_override(address, connect_timeout).await?
                 };
 
                 // replace socks user_id with config.selected_scope, if set
@@ -449,16 +631,32 @@ impl UpstreamManager {
                     .filter(|s| !s.is_empty());
                 let _user_id: Option<&str> = scope.or(user_id.as_deref());
 
-                match tokio::time::timeout(connect_timeout, connect_socks4(&mut stream, target, _user_id)).await {
-                    Ok(Ok(())) => {}
+                let bound = match tokio::time::timeout(
+                    connect_timeout,
+                    connect_socks4(&mut stream, target, _user_id),
+                )
+                .await
+                {
+                    Ok(Ok(bound)) => bound,
                     Ok(Err(e)) => return Err(e),
                     Err(_) => {
                         return Err(ProxyError::ConnectionTimeout {
                             addr: target.to_string(),
                         });
                     }
-                }
-                Ok(stream)
+                };
+                let local_addr = stream.local_addr().ok();
+                let socks_proxy_addr = stream.peer_addr().ok();
+                Ok((
+                    stream,
+                    UpstreamEgressInfo {
+                        route_kind: UpstreamRouteKind::Socks4,
+                        local_addr,
+                        direct_bind_ip: None,
+                        socks_bound_addr: Some(bound.addr),
+                        socks_proxy_addr,
+                    },
+                ))
             },
             UpstreamType::Socks5 { address, interface, username, password } => {
                 let connect_timeout = Duration::from_secs(DIRECT_CONNECT_TIMEOUT_SECS);
@@ -470,6 +668,7 @@ impl UpstreamManager {
                         &None,
                         proxy_addr,
                         bind_rr.as_deref(),
+                        false,
                     );
 
                     let socket = create_outgoing_socket_bound(proxy_addr, bind_ip)?;
@@ -503,15 +702,7 @@ impl UpstreamManager {
                     if interface.is_some() {
                         warn!("SOCKS5 interface binding is not supported for hostname addresses, ignoring");
                     }
-                    match tokio::time::timeout(connect_timeout, TcpStream::connect(address)).await {
-                        Ok(Ok(stream)) => stream,
-                        Ok(Err(e)) => return Err(ProxyError::Io(e)),
-                        Err(_) => {
-                            return Err(ProxyError::ConnectionTimeout {
-                                addr: address.clone(),
-                            });
-                        }
-                    }
+                    Self::connect_hostname_with_dns_override(address, connect_timeout).await?
                 };
 
                 debug!(config = ?config, "Socks5 connection");
@@ -521,21 +712,32 @@ impl UpstreamManager {
                 let _username: Option<&str> = scope.or(username.as_deref());
                 let _password: Option<&str> = scope.or(password.as_deref());
 
-                match tokio::time::timeout(
+                let bound = match tokio::time::timeout(
                     connect_timeout,
                     connect_socks5(&mut stream, target, _username, _password),
                 )
                 .await
                 {
-                    Ok(Ok(())) => {}
+                    Ok(Ok(bound)) => bound,
                     Ok(Err(e)) => return Err(e),
                     Err(_) => {
                         return Err(ProxyError::ConnectionTimeout {
                             addr: target.to_string(),
                         });
                     }
-                }
-                Ok(stream)
+                };
+                let local_addr = stream.local_addr().ok();
+                let socks_proxy_addr = stream.peer_addr().ok();
+                Ok((
+                    stream,
+                    UpstreamEgressInfo {
+                        route_kind: UpstreamRouteKind::Socks5,
+                        local_addr,
+                        direct_bind_ip: None,
+                        socks_bound_addr: Some(bound.addr),
+                        socks_proxy_addr,
+                    },
+                ))
             },
         }
     }
@@ -562,8 +764,22 @@ impl UpstreamManager {
 
         for (upstream_idx, upstream_config, bind_rr) in &upstreams {
             let upstream_name = match &upstream_config.upstream_type {
-                UpstreamType::Direct { interface, .. } => {
-                    format!("direct{}", interface.as_ref().map(|i| format!(" ({})", i)).unwrap_or_default())
+                UpstreamType::Direct {
+                    interface,
+                    bind_addresses,
+                } => {
+                    let mut direct_parts = Vec::new();
+                    if let Some(dev) = interface.as_deref().filter(|v| !v.is_empty()) {
+                        direct_parts.push(format!("dev={dev}"));
+                    }
+                    if let Some(src) = bind_addresses.as_ref().filter(|v| !v.is_empty()) {
+                        direct_parts.push(format!("src={}", src.join(",")));
+                    }
+                    if direct_parts.is_empty() {
+                        "direct".to_string()
+                    } else {
+                        format!("direct {}", direct_parts.join(" "))
+                    }
                 }
                 UpstreamType::Socks4 { address, .. } => format!("socks4://{}", address),
                 UpstreamType::Socks5 { address, .. } => format!("socks5://{}", address),
@@ -767,7 +983,7 @@ impl UpstreamManager {
         target: SocketAddr,
     ) -> Result<f64> {
         let start = Instant::now();
-        let _stream = self.connect_via_upstream(config, target, bind_rr).await?;
+        let _ = self.connect_via_upstream(config, target, bind_rr).await?;
         Ok(start.elapsed().as_secs_f64() * 1000.0)
     }
 
@@ -870,18 +1086,26 @@ impl UpstreamManager {
                                     u.fails += 1;
                                     debug!(dc = dc_zero_idx + 1, fails = u.fails,
                                         "Health check failed (both): {}", e);
-                                    if u.fails > 3 {
+                                    if u.fails >= self.unhealthy_fail_threshold {
                                         u.healthy = false;
-                                        warn!("Upstream unhealthy (fails)");
+                                        warn!(
+                                            fails = u.fails,
+                                            threshold = self.unhealthy_fail_threshold,
+                                            "Upstream unhealthy (fails)"
+                                        );
                                     }
                                 }
                                 Err(_) => {
                                     u.fails += 1;
                                     debug!(dc = dc_zero_idx + 1, fails = u.fails,
                                         "Health check timeout (both)");
-                                    if u.fails > 3 {
+                                    if u.fails >= self.unhealthy_fail_threshold {
                                         u.healthy = false;
-                                        warn!("Upstream unhealthy (timeout)");
+                                        warn!(
+                                            fails = u.fails,
+                                            threshold = self.unhealthy_fail_threshold,
+                                            "Upstream unhealthy (timeout)"
+                                        );
                                     }
                                 }
                             }
@@ -892,9 +1116,13 @@ impl UpstreamManager {
                         let mut guard = self.upstreams.write().await;
                         let u = &mut guard[i];
                         u.fails += 1;
-                        if u.fails > 3 {
+                        if u.fails >= self.unhealthy_fail_threshold {
                             u.healthy = false;
-                            warn!("Upstream unhealthy (no fallback family)");
+                            warn!(
+                                fails = u.fails,
+                                threshold = self.unhealthy_fail_threshold,
+                                "Upstream unhealthy (no fallback family)"
+                            );
                         }
                         u.last_check = std::time::Instant::now();
                     }