Update release.yml

Update release.yml

.dockerignore

@@ -0,0 +1,8 @@
+.git
+.github
+target
+.kilocode
+cache
+tlsfront
+*.tar
+*.tar.gz

.github/workflows/release.yml

@@ -6,36 +6,34 @@ on:
       - '[0-9]+.[0-9]+.[0-9]+'
   workflow_dispatch:
 
+concurrency:
+  group: release-${{ github.ref }}
+  cancel-in-progress: true
+
 permissions:
   contents: read
   packages: write
 
 env:
   CARGO_TERM_COLOR: always
+  BINARY_NAME: telemt
 
 jobs:
-  build:
+# ==========================
-    name: Build ${{ matrix.target }}
+# GNU / glibc
+# ==========================
+  build-gnu:
+    name: GNU ${{ matrix.target }}
     runs-on: ubuntu-latest
-    permissions:
-      contents: read
 
     strategy:
       fail-fast: false
       matrix:
         include:
           - target: x86_64-unknown-linux-gnu
-            artifact_name: telemt
+            asset: telemt-x86_64-linux-gnu
-            asset_name: telemt-x86_64-linux-gnu
           - target: aarch64-unknown-linux-gnu
-            artifact_name: telemt
+            asset: telemt-aarch64-linux-gnu
-            asset_name: telemt-aarch64-linux-gnu
-          - target: x86_64-unknown-linux-musl
-            artifact_name: telemt
-            asset_name: telemt-x86_64-linux-musl
-          - target: aarch64-unknown-linux-musl
-            artifact_name: telemt
-            asset_name: telemt-aarch64-linux-musl
 
     steps:
       - uses: actions/checkout@v4
@@ -43,12 +41,20 @@ jobs:
       - uses: dtolnay/rust-toolchain@v1
         with:
           toolchain: stable
-          targets: ${{ matrix.target }}
+          targets: |
+            x86_64-unknown-linux-gnu
+            aarch64-unknown-linux-gnu
 
-      - name: Install cross-compilation tools
+      - name: Install deps
         run: |
           sudo apt-get update
-          sudo apt-get install -y gcc-aarch64-linux-gnu
+          sudo apt-get install -y \
+            build-essential \
+            clang \
+            lld \
+            pkg-config \
+            gcc-aarch64-linux-gnu \
+            g++-aarch64-linux-gnu
 
       - uses: actions/cache@v4
         with:
@@ -56,41 +62,183 @@ jobs:
             ~/.cargo/registry
             ~/.cargo/git
             target
-          key: ${{ runner.os }}-${{ matrix.target }}-cargo-${{ hashFiles('**/Cargo.lock') }}
+          key: gnu-${{ matrix.target }}-${{ hashFiles('**/Cargo.lock') }}
-          restore-keys: |
-            ${{ runner.os }}-${{ matrix.target }}-cargo-
 
-      - name: Install cross
+      - name: Build
-        run: cargo install cross --git https://github.com/cross-rs/cross
-
-      - name: Build Release
-        env:
-          RUSTFLAGS: ${{ contains(matrix.target, 'musl') && '-C target-feature=+crt-static' || '' }}
-        run: cross build --release --target ${{ matrix.target }}
-
-      - name: Package binary
         run: |
-          cd target/${{ matrix.target }}/release
+          if [ "${{ matrix.target }}" = "aarch64-unknown-linux-gnu" ]; then
-          tar -czvf ${{ matrix.asset_name }}.tar.gz ${{ matrix.artifact_name }}
+            export CC=aarch64-linux-gnu-gcc
-          sha256sum ${{ matrix.asset_name }}.tar.gz > ${{ matrix.asset_name }}.sha256
+            export CXX=aarch64-linux-gnu-g++
+            export CC_aarch64_unknown_linux_gnu=aarch64-linux-gnu-gcc
+            export CXX_aarch64_unknown_linux_gnu=aarch64-linux-gnu-g++
+            export RUSTFLAGS="-C linker=aarch64-linux-gnu-gcc"
+          else
+            export CC=clang
+            export CXX=clang++
+            export CC_x86_64_unknown_linux_gnu=clang
+            export CXX_x86_64_unknown_linux_gnu=clang++
+            export RUSTFLAGS="-C linker=clang -C link-arg=-fuse-ld=lld"
+          fi
+
+          cargo build --release --target ${{ matrix.target }}
+
+      - name: Package
+        run: |
+          mkdir -p dist
+          BIN=target/${{ matrix.target }}/release/${{ env.BINARY_NAME }}
+
+          cp "$BIN" dist/${{ env.BINARY_NAME }}-${{ matrix.target }}
+
+          cd dist
+          tar -czf ${{ matrix.asset }}.tar.gz ${{ env.BINARY_NAME }}-${{ matrix.target }}
+          sha256sum ${{ matrix.asset }}.tar.gz > ${{ matrix.asset }}.sha256
 
       - uses: actions/upload-artifact@v4
         with:
-          name: ${{ matrix.asset_name }}
+          name: ${{ matrix.asset }}
           path: |
-            target/${{ matrix.target }}/release/${{ matrix.asset_name }}.tar.gz
+            dist/${{ matrix.asset }}.tar.gz
-            target/${{ matrix.target }}/release/${{ matrix.asset_name }}.sha256
+            dist/${{ matrix.asset }}.sha256
 
-  build-docker-image:
+# ==========================
-    needs: build
+# MUSL
+# ==========================
+  build-musl:
+    name: MUSL ${{ matrix.target }}
     runs-on: ubuntu-latest
-    permissions:
+
-      contents: read
+    container:
-      packages: write
+      image: rust:slim-bookworm
+
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - target: x86_64-unknown-linux-musl
+            asset: telemt-x86_64-linux-musl
+          - target: aarch64-unknown-linux-musl
+            asset: telemt-aarch64-linux-musl
 
     steps:
       - uses: actions/checkout@v4
 
+      - name: Install deps
+        run: |
+          apt-get update
+          apt-get install -y \
+            musl-tools \
+            pkg-config \
+            curl
+
+      # 💾 cache toolchain
+      - uses: actions/cache@v4
+        if: matrix.target == 'aarch64-unknown-linux-musl'
+        with:
+          path: ~/.musl-aarch64
+          key: musl-toolchain-aarch64-v1
+
+      # 🔥 надёжная установка
+      - name: Install aarch64 musl toolchain
+        if: matrix.target == 'aarch64-unknown-linux-musl'
+        run: |
+          set -e
+
+          TOOLCHAIN_DIR="$HOME/.musl-aarch64"
+          ARCHIVE="aarch64-linux-musl-cross.tgz"
+
+          if [ -x "$TOOLCHAIN_DIR/bin/aarch64-linux-musl-gcc" ]; then
+            echo "✅ musl toolchain already installed"
+          else
+            echo "⬇️ downloading musl toolchain..."
+
+            download() {
+              url="$1"
+              echo "→ trying $url"
+              curl -fL \
+                --retry 5 \
+                --retry-delay 3 \
+                --connect-timeout 10 \
+                --max-time 120 \
+                -o "$ARCHIVE" "$url" && return 0
+              return 1
+            }
+
+            download "https://musl.cc/$ARCHIVE" || \
+            download "https://more.musl.cc/$ARCHIVE" || \
+            { echo "❌ failed to download musl toolchain"; exit 1; }
+
+            mkdir -p "$TOOLCHAIN_DIR"
+            tar -xzf "$ARCHIVE" --strip-components=1 -C "$TOOLCHAIN_DIR"
+          fi
+
+          echo "$TOOLCHAIN_DIR/bin" >> $GITHUB_PATH
+
+      - name: Add rust target
+        run: rustup target add ${{ matrix.target }}
+
+      - uses: actions/cache@v4
+        with:
+          path: |
+            /usr/local/cargo/registry
+            /usr/local/cargo/git
+            target
+          key: musl-${{ matrix.target }}-${{ hashFiles('**/Cargo.lock') }}
+
+      - name: Build
+        run: |
+          if [ "${{ matrix.target }}" = "aarch64-unknown-linux-musl" ]; then
+            export CC=aarch64-linux-musl-gcc
+            export CC_aarch64_unknown_linux_musl=aarch64-linux-musl-gcc
+            export RUSTFLAGS="-C target-feature=+crt-static -C linker=aarch64-linux-musl-gcc"
+          else
+            export CC=musl-gcc
+            export CC_x86_64_unknown_linux_musl=musl-gcc
+            export RUSTFLAGS="-C target-feature=+crt-static"
+          fi
+
+          cargo build --release --target ${{ matrix.target }}
+
+      - name: Package
+        run: |
+          mkdir -p dist
+          BIN=target/${{ matrix.target }}/release/${{ env.BINARY_NAME }}
+
+          cp "$BIN" dist/${{ env.BINARY_NAME }}-${{ matrix.target }}
+
+          cd dist
+          tar -czf ${{ matrix.asset }}.tar.gz ${{ env.BINARY_NAME }}-${{ matrix.target }}
+          sha256sum ${{ matrix.asset }}.tar.gz > ${{ matrix.asset }}.sha256
+
+      - uses: actions/upload-artifact@v4
+        with:
+          name: ${{ matrix.asset }}
+          path: |
+            dist/${{ matrix.asset }}.tar.gz
+            dist/${{ matrix.asset }}.sha256
+
+# ==========================
+# Docker
+# ==========================
+  docker:
+    name: Docker
+    runs-on: ubuntu-latest
+    needs: [build-gnu, build-musl]
+    continue-on-error: true
+
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: actions/download-artifact@v4
+        with:
+          path: artifacts
+
+      - name: Extract binaries
+        run: |
+          mkdir dist
+          find artifacts -name "*.tar.gz" -exec tar -xzf {} -C dist \;
+
+          cp dist/telemt-x86_64-unknown-linux-musl dist/telemt || true
+
       - uses: docker/setup-qemu-action@v3
       - uses: docker/setup-buildx-action@v3
 
@@ -105,35 +253,43 @@ jobs:
         id: vars
         run: echo "VERSION=${GITHUB_REF#refs/tags/}" >> $GITHUB_OUTPUT
 
-      - name: Build and push
+      - name: Build & Push
         uses: docker/build-push-action@v6
         with:
           context: .
           push: true
+          platforms: linux/amd64,linux/arm64
           tags: |
             ghcr.io/${{ github.repository }}:${{ steps.vars.outputs.VERSION }}
             ghcr.io/${{ github.repository }}:latest
+          build-args: |
+            BINARY=dist/telemt
 
+# ==========================
+# Release
+# ==========================
   release:
-    name: Create Release
+    name: Release
-    needs: build
     runs-on: ubuntu-latest
+    needs: [build-gnu, build-musl]
+
     permissions:
       contents: write
 
     steps:
-      - uses: actions/checkout@v4
-        with:
-          fetch-depth: 0
-
       - uses: actions/download-artifact@v4
         with:
           path: artifacts
 
+      - name: Flatten artifacts
+        run: |
+          mkdir dist
+          find artifacts -type f -exec cp {} dist/ \;
+
       - name: Create Release
         uses: softprops/action-gh-release@v2
         with:
-          files: artifacts/**/*
+          files: dist/*
           generate_release_notes: true
           draft: false
           prerelease: ${{ contains(github.ref, '-rc') || contains(github.ref, '-beta') || contains(github.ref, '-alpha') }}

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "telemt"
-version = "3.3.25"
+version = "3.3.28"
 edition = "2024"
 
 [dependencies]
@@ -26,6 +26,7 @@ zeroize = { version = "1.8", features = ["derive"] }
 # Network
 socket2 = { version = "0.5", features = ["all"] }
 nix = { version = "0.28", default-features = false, features = ["net"] }
+shadowsocks = { version = "1.24", features = ["aead-cipher-2022"] }
 
 # Serialization
 serde = { version = "1.0", features = ["derive"] }
@@ -40,6 +41,7 @@ tracing = "0.1"
 tracing-subscriber = { version = "0.3", features = ["env-filter"] }
 parking_lot = "0.12"
 dashmap = "5.5"
+arc-swap = "1.7"
 lru = "0.16"
 rand = "0.9"
 chrono = { version = "0.4", features = ["serde"] }

Dockerfile

@@ -1,3 +1,5 @@
+# syntax=docker/dockerfile:1
+
 # ==========================
 # Stage 1: Build
 # ==========================
@@ -5,36 +7,87 @@ FROM rust:1.88-slim-bookworm AS builder
 
 RUN apt-get update && apt-get install -y --no-install-recommends \
     pkg-config \
+    ca-certificates \
     && rm -rf /var/lib/apt/lists/*
 
 WORKDIR /build
 
+# Depcache
 COPY Cargo.toml Cargo.lock* ./
 RUN mkdir src && echo 'fn main() {}' > src/main.rs && \
     cargo build --release 2>/dev/null || true && \
     rm -rf src
 
+# Build
 COPY . .
 RUN cargo build --release && strip target/release/telemt
 
 # ==========================
-# Stage 2: Runtime
+# Stage 2: Compress (strip + UPX)
 # ==========================
-FROM debian:bookworm-slim
+FROM debian:12-slim AS minimal
+
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    upx \
+    binutils \
+    && rm -rf /var/lib/apt/lists/*
+
+COPY --from=builder /build/target/release/telemt /telemt
+
+RUN strip /telemt || true
+RUN upx --best --lzma /telemt || true
+
+# ==========================
+# Stage 3: Debug base
+# ==========================
+FROM debian:12-slim AS debug-base
 
 RUN apt-get update && apt-get install -y --no-install-recommends \
     ca-certificates \
+    tzdata \
+    curl \
+    iproute2 \
+    busybox \
     && rm -rf /var/lib/apt/lists/*
 
-RUN useradd -r -s /usr/sbin/nologin telemt
+# ==========================
+# Stage 4: Debug image
+# ==========================
+FROM debug-base AS debug
 
 WORKDIR /app
 
-COPY --from=builder /build/target/release/telemt /app/telemt
+COPY --from=minimal /telemt /app/telemt
 COPY config.toml /app/config.toml
 
-RUN chown -R telemt:telemt /app
+USER root
-USER telemt
+
+EXPOSE 443
+EXPOSE 9090
+EXPOSE 9091
+
+ENTRYPOINT ["/app/telemt"]
+CMD ["config.toml"]
+
+# ==========================
+# Stage 5: Production (distroless)
+# ==========================
+FROM gcr.io/distroless/base-debian12 AS prod
+
+WORKDIR /app
+
+COPY --from=minimal /telemt /app/telemt
+COPY config.toml /app/config.toml
+
+# TLS + timezone + shell
+COPY --from=debug-base /etc/ssl/certs /etc/ssl/certs
+COPY --from=debug-base /usr/share/zoneinfo /usr/share/zoneinfo
+COPY --from=debug-base /bin/busybox /bin/busybox
+
+RUN ["/bin/busybox", "--install", "-s", "/bin"]
+
+# distroless user
+USER nonroot:nonroot
 
 EXPOSE 443
 EXPOSE 9090

README.md

@@ -19,9 +19,9 @@
 
 ### 🇷🇺 RU
 
-#### Релиз 3.3.15 Semistable
+#### О релизах
 
-[3.3.15](https://github.com/telemt/telemt/releases/tag/3.3.15) по итогам работы в продакшн признан одним из самых стабильных и рекомендуется к использованию, когда cutting-edge фичи некритичны!
+[3.3.27](https://github.com/telemt/telemt/releases/tag/3.3.27) даёт баланс стабильности и передового функционала, а так же последние исправления по безопасности и багам
 
 Будем рады вашему фидбеку и предложениям по улучшению — особенно в части **API**, **статистики**, **UX**
 
@@ -40,9 +40,9 @@
 
 ### 🇬🇧 EN
 
-#### Release 3.3.15 Semistable
+#### About releases
 
-[3.3.15](https://github.com/telemt/telemt/releases/tag/3.3.15) is, based on the results of his work in production, recognized as one of the most stable and recommended for use when cutting-edge features are not so necessary!
+[3.3.27](https://github.com/telemt/telemt/releases/tag/3.3.27) provides a balance of stability and advanced functionality, as well as the latest security and bug fixes
 
 We are looking forward to your feedback and improvement proposals — especially regarding **API**, **statistics**, **UX**
 

docs/API.md

@@ -497,13 +497,14 @@ Note: the request contract is defined, but the corresponding route currently ret
 | `direct_total` | `usize` | Direct-route upstream entries. |
 | `socks4_total` | `usize` | SOCKS4 upstream entries. |
 | `socks5_total` | `usize` | SOCKS5 upstream entries. |
+| `shadowsocks_total` | `usize` | Shadowsocks upstream entries. |
 
 #### `RuntimeUpstreamQualityUpstreamData`
 | Field | Type | Description |
 | --- | --- | --- |
 | `upstream_id` | `usize` | Runtime upstream index. |
-| `route_kind` | `string` | `direct`, `socks4`, `socks5`. |
+| `route_kind` | `string` | `direct`, `socks4`, `socks5`, `shadowsocks`. |
-| `address` | `string` | Upstream address (`direct` literal for direct route kind). |
+| `address` | `string` | Upstream address (`direct` literal for direct route kind, `host:port` only for proxied upstreams). |
 | `weight` | `u16` | Selection weight. |
 | `scopes` | `string` | Configured scope selector. |
 | `healthy` | `bool` | Current health flag. |
@@ -757,13 +758,14 @@ Note: the request contract is defined, but the corresponding route currently ret
 | `direct_total` | `usize` | Number of direct upstream entries. |
 | `socks4_total` | `usize` | Number of SOCKS4 upstream entries. |
 | `socks5_total` | `usize` | Number of SOCKS5 upstream entries. |
+| `shadowsocks_total` | `usize` | Number of Shadowsocks upstream entries. |
 
 #### `UpstreamStatus`
 | Field | Type | Description |
 | --- | --- | --- |
 | `upstream_id` | `usize` | Runtime upstream index. |
-| `route_kind` | `string` | Upstream route kind: `direct`, `socks4`, `socks5`. |
+| `route_kind` | `string` | Upstream route kind: `direct`, `socks4`, `socks5`, `shadowsocks`. |
-| `address` | `string` | Upstream address (`direct` for direct route kind). Authentication fields are intentionally omitted. |
+| `address` | `string` | Upstream address (`direct` for direct route kind, `host:port` for Shadowsocks). Authentication fields are intentionally omitted. |
 | `weight` | `u16` | Selection weight. |
 | `scopes` | `string` | Configured scope selector string. |
 | `healthy` | `bool` | Current health flag. |

docs/FAQ.en.md

@@ -120,3 +120,17 @@ password = "pass"          # Password for Auth on SOCKS-server
 weight = 1                 # Set Weight for Scenarios
 enabled = true
 ```
+
+#### Shadowsocks as Upstream
+Requires `use_middle_proxy = false`.
+
+```toml
+[general]
+use_middle_proxy = false
+
+[[upstreams]]
+type = "shadowsocks"
+url = "ss://2022-blake3-aes-256-gcm:BASE64_KEY@1.2.3.4:8388"
+weight = 1
+enabled = true
+```

docs/FAQ.ru.md

@@ -121,3 +121,16 @@ weight = 1                 # Set Weight for Scenarios
 enabled = true
 ```
 
+#### Shadowsocks как Upstream
+Требует `use_middle_proxy = false`.
+
+```toml
+[general]
+use_middle_proxy = false
+
+[[upstreams]]
+type = "shadowsocks"
+url = "ss://2022-blake3-aes-256-gcm:BASE64_KEY@1.2.3.4:8388"
+weight = 1
+enabled = true
+```

docs/QUICK_START_GUIDE.ru.md

@@ -178,7 +178,7 @@ docker compose down
 > - По умолчанию публикуются порты 443:443, а контейнер запускается со сброшенными привилегиями (добавлена только `NET_BIND_SERVICE`)  
 > - Если вам действительно нужна сеть хоста (обычно это требуется только для некоторых конфигураций IPv6), раскомментируйте `network_mode: host`
 
-**Запуск в Docker Compose**
+**Запуск без Docker Compose**
 ```bash
 docker build -t telemt:local .
 docker run --name telemt --restart unless-stopped \

docs/TUNING.de.md

@@ -82,7 +82,7 @@ Die unten angegebenen `Default`-Werte sind Code-Defaults (bei fehlendem Schlüss
 
 | Feld | Gilt für | Typ | Pflicht | Default | Bedeutung |
 |---|---|---|---|---|---|
-| `[[upstreams]].type` | alle Upstreams | `"direct" \| "socks4" \| "socks5"` | ja | n/a | Upstream-Transporttyp. |
+| `[[upstreams]].type` | alle Upstreams | `"direct" \| "socks4" \| "socks5" \| "shadowsocks"` | 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. |
@@ -95,6 +95,8 @@ Die unten angegebenen `Default`-Werte sind Code-Defaults (bei fehlendem Schlüss
 | `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. |
+| `url` | `shadowsocks` | `String` | ja | n/a | Shadowsocks-SIP002-URL (`ss://...`). In Runtime-APIs wird nur `host:port` offengelegt. |
+| `interface` | `shadowsocks` | `Option<String>` | nein | `null` | Optionales ausgehendes Bind-Interface oder lokale Literal-IP. |
 
 ### Runtime-Regeln (wichtig)
 
@@ -115,6 +117,7 @@ Die unten angegebenen `Default`-Werte sind Code-Defaults (bei fehlendem Schlüss
 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.
+11. `shadowsocks`-Upstreams erfordern `general.use_middle_proxy = false`. Mit aktiviertem ME-Modus schlägt das Laden der Config sofort fehl.
 
 ## Upstream-Konfigurationsbeispiele
 
@@ -150,7 +153,20 @@ weight = 2
 enabled = true
 ```
 
-### Beispiel 4: Gemischte Upstreams mit Scopes
+### Beispiel 4: Shadowsocks-Upstream
+
+```toml
+[general]
+use_middle_proxy = false
+
+[[upstreams]]
+type = "shadowsocks"
+url = "ss://2022-blake3-aes-256-gcm:BASE64_KEY@198.51.100.50:8388"
+weight = 2
+enabled = true
+```
+
+### Beispiel 5: Gemischte Upstreams mit Scopes
 
 ```toml
 [[upstreams]]

docs/TUNING.en.md

@@ -82,7 +82,7 @@ Defaults below are code defaults (used when a key is omitted), not necessarily v
 
 | Field | Applies to | Type | Required | Default | Meaning |
 |---|---|---|---|---|---|
-| `[[upstreams]].type` | all upstreams | `"direct" \| "socks4" \| "socks5"` | yes | n/a | Upstream transport type. |
+| `[[upstreams]].type` | all upstreams | `"direct" \| "socks4" \| "socks5" \| "shadowsocks"` | 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. |
@@ -95,6 +95,8 @@ Defaults below are code defaults (used when a key is omitted), not necessarily v
 | `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. |
+| `url` | `shadowsocks` | `String` | yes | n/a | Shadowsocks SIP002 URL (`ss://...`). Only `host:port` is exposed in runtime APIs. |
+| `interface` | `shadowsocks` | `Option<String>` | no | `null` | Optional outgoing bind interface or literal local IP. |
 
 ### Runtime rules (important)
 
@@ -115,6 +117,7 @@ Defaults below are code defaults (used when a key is omitted), not necessarily v
 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.
+11. `shadowsocks` upstreams require `general.use_middle_proxy = false`. Config load fails fast if ME mode is enabled.
 
 ## Upstream Configuration Examples
 
@@ -150,7 +153,20 @@ weight = 2
 enabled = true
 ```
 
-### Example 4: Mixed upstreams with scopes
+### Example 4: Shadowsocks upstream
+
+```toml
+[general]
+use_middle_proxy = false
+
+[[upstreams]]
+type = "shadowsocks"
+url = "ss://2022-blake3-aes-256-gcm:BASE64_KEY@198.51.100.50:8388"
+weight = 2
+enabled = true
+```
+
+### Example 5: Mixed upstreams with scopes
 
 ```toml
 [[upstreams]]

docs/TUNING.ru.md

@@ -82,7 +82,7 @@
 
 | Поле | Применимость | Тип | Обязательно | Default | Назначение |
 |---|---|---|---|---|---|
-| `[[upstreams]].type` | все upstream | `"direct" \| "socks4" \| "socks5"` | да | n/a | Тип upstream транспорта. |
+| `[[upstreams]].type` | все upstream | `"direct" \| "socks4" \| "socks5" \| "shadowsocks"` | 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. |
@@ -95,6 +95,8 @@
 | `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. |
+| `url` | `shadowsocks` | `String` | да | n/a | Shadowsocks SIP002 URL (`ss://...`). В runtime API раскрывается только `host:port`. |
+| `interface` | `shadowsocks` | `Option<String>` | нет | `null` | Необязательный исходящий bind-интерфейс или literal локальный IP. |
 
 ### Runtime-правила
 
@@ -115,6 +117,7 @@
 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.
+11. `shadowsocks` upstream требует `general.use_middle_proxy = false`. При включенном ME-режиме конфиг отклоняется при загрузке.
 
 ## Примеры конфигурации Upstreams
 
@@ -150,7 +153,20 @@ weight = 2
 enabled = true
 ```
 
-### Пример 4: смешанные upstream с scopes
+### Пример 4: Shadowsocks upstream
+
+```toml
+[general]
+use_middle_proxy = false
+
+[[upstreams]]
+type = "shadowsocks"
+url = "ss://2022-blake3-aes-256-gcm:BASE64_KEY@198.51.100.50:8388"
+weight = 2
+enabled = true
+```
+
+### Пример 5: смешанные upstream с scopes
 
 ```toml
 [[upstreams]]

src/api/model.rs

@@ -134,6 +134,7 @@ pub(super) struct UpstreamSummaryData {
     pub(super) direct_total: usize,
     pub(super) socks4_total: usize,
     pub(super) socks5_total: usize,
+    pub(super) shadowsocks_total: usize,
 }
 
 #[derive(Serialize, Clone)]
@@ -205,6 +206,16 @@ pub(super) struct ZeroPoolData {
     pub(super) refill_failed_total: u64,
     pub(super) writer_restored_same_endpoint_total: u64,
     pub(super) writer_restored_fallback_total: u64,
+    pub(super) teardown_attempt_total_normal: u64,
+    pub(super) teardown_attempt_total_hard_detach: u64,
+    pub(super) teardown_success_total_normal: u64,
+    pub(super) teardown_success_total_hard_detach: u64,
+    pub(super) teardown_timeout_total: u64,
+    pub(super) teardown_escalation_total: u64,
+    pub(super) teardown_noop_total: u64,
+    pub(super) teardown_cleanup_side_effect_failures_total: u64,
+    pub(super) teardown_duration_count_total: u64,
+    pub(super) teardown_duration_sum_seconds_total: f64,
 }
 
 #[derive(Serialize, Clone)]

src/api/runtime_min.rs

@@ -4,6 +4,9 @@ use std::time::{SystemTime, UNIX_EPOCH};
 use serde::Serialize;
 
 use crate::config::ProxyConfig;
+use crate::stats::{
+    MeWriterCleanupSideEffectStep, MeWriterTeardownMode, MeWriterTeardownReason, Stats,
+};
 
 use super::ApiShared;
 
@@ -98,6 +101,50 @@ pub(super) struct RuntimeMeQualityCountersData {
     pub(super) reconnect_success_total: u64,
 }
 
+#[derive(Serialize)]
+pub(super) struct RuntimeMeQualityTeardownAttemptData {
+    pub(super) reason: &'static str,
+    pub(super) mode: &'static str,
+    pub(super) total: u64,
+}
+
+#[derive(Serialize)]
+pub(super) struct RuntimeMeQualityTeardownSuccessData {
+    pub(super) mode: &'static str,
+    pub(super) total: u64,
+}
+
+#[derive(Serialize)]
+pub(super) struct RuntimeMeQualityTeardownSideEffectData {
+    pub(super) step: &'static str,
+    pub(super) total: u64,
+}
+
+#[derive(Serialize)]
+pub(super) struct RuntimeMeQualityTeardownDurationBucketData {
+    pub(super) le_seconds: &'static str,
+    pub(super) total: u64,
+}
+
+#[derive(Serialize)]
+pub(super) struct RuntimeMeQualityTeardownDurationData {
+    pub(super) mode: &'static str,
+    pub(super) count: u64,
+    pub(super) sum_seconds: f64,
+    pub(super) buckets: Vec<RuntimeMeQualityTeardownDurationBucketData>,
+}
+
+#[derive(Serialize)]
+pub(super) struct RuntimeMeQualityTeardownData {
+    pub(super) attempts: Vec<RuntimeMeQualityTeardownAttemptData>,
+    pub(super) success: Vec<RuntimeMeQualityTeardownSuccessData>,
+    pub(super) timeout_total: u64,
+    pub(super) escalation_total: u64,
+    pub(super) noop_total: u64,
+    pub(super) cleanup_side_effect_failures: Vec<RuntimeMeQualityTeardownSideEffectData>,
+    pub(super) duration: Vec<RuntimeMeQualityTeardownDurationData>,
+}
+
 #[derive(Serialize)]
 pub(super) struct RuntimeMeQualityRouteDropData {
     pub(super) no_conn_total: u64,
@@ -107,6 +154,25 @@ pub(super) struct RuntimeMeQualityRouteDropData {
     pub(super) queue_full_high_total: u64,
 }
 
+#[derive(Serialize)]
+pub(super) struct RuntimeMeQualityFamilyStateData {
+    pub(super) family: &'static str,
+    pub(super) state: &'static str,
+    pub(super) state_since_epoch_secs: u64,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub(super) suppressed_until_epoch_secs: Option<u64>,
+    pub(super) fail_streak: u32,
+    pub(super) recover_success_streak: u32,
+}
+
+#[derive(Serialize)]
+pub(super) struct RuntimeMeQualityDrainGateData {
+    pub(super) route_quorum_ok: bool,
+    pub(super) redundancy_ok: bool,
+    pub(super) block_reason: &'static str,
+    pub(super) updated_at_epoch_secs: u64,
+}
+
 #[derive(Serialize)]
 pub(super) struct RuntimeMeQualityDcRttData {
     pub(super) dc: i16,
@@ -120,7 +186,10 @@ pub(super) struct RuntimeMeQualityDcRttData {
 #[derive(Serialize)]
 pub(super) struct RuntimeMeQualityPayload {
     pub(super) counters: RuntimeMeQualityCountersData,
+    pub(super) teardown: RuntimeMeQualityTeardownData,
     pub(super) route_drops: RuntimeMeQualityRouteDropData,
+    pub(super) family_states: Vec<RuntimeMeQualityFamilyStateData>,
+    pub(super) drain_gate: RuntimeMeQualityDrainGateData,
     pub(super) dc_rtt: Vec<RuntimeMeQualityDcRttData>,
 }
 
@@ -159,6 +228,7 @@ pub(super) struct RuntimeUpstreamQualitySummaryData {
     pub(super) direct_total: usize,
     pub(super) socks4_total: usize,
     pub(super) socks5_total: usize,
+    pub(super) shadowsocks_total: usize,
 }
 
 #[derive(Serialize)]
@@ -361,6 +431,19 @@ pub(super) async fn build_runtime_me_quality_data(shared: &ApiShared) -> Runtime
     };
 
     let status = pool.api_status_snapshot().await;
+    let family_states = pool
+        .api_family_state_snapshot()
+        .into_iter()
+        .map(|entry| RuntimeMeQualityFamilyStateData {
+            family: entry.family,
+            state: entry.state,
+            state_since_epoch_secs: entry.state_since_epoch_secs,
+            suppressed_until_epoch_secs: entry.suppressed_until_epoch_secs,
+            fail_streak: entry.fail_streak,
+            recover_success_streak: entry.recover_success_streak,
+        })
+        .collect();
+    let drain_gate_snapshot = pool.api_drain_gate_snapshot();
     RuntimeMeQualityData {
         enabled: true,
         reason: None,
@@ -374,6 +457,7 @@ pub(super) async fn build_runtime_me_quality_data(shared: &ApiShared) -> Runtime
                 reconnect_attempt_total: shared.stats.get_me_reconnect_attempts(),
                 reconnect_success_total: shared.stats.get_me_reconnect_success(),
             },
+            teardown: build_runtime_me_teardown_data(shared),
             route_drops: RuntimeMeQualityRouteDropData {
                 no_conn_total: shared.stats.get_me_route_drop_no_conn(),
                 channel_closed_total: shared.stats.get_me_route_drop_channel_closed(),
@@ -381,6 +465,13 @@ pub(super) async fn build_runtime_me_quality_data(shared: &ApiShared) -> Runtime
                 queue_full_base_total: shared.stats.get_me_route_drop_queue_full_base(),
                 queue_full_high_total: shared.stats.get_me_route_drop_queue_full_high(),
             },
+            family_states,
+            drain_gate: RuntimeMeQualityDrainGateData {
+                route_quorum_ok: drain_gate_snapshot.route_quorum_ok,
+                redundancy_ok: drain_gate_snapshot.redundancy_ok,
+                block_reason: drain_gate_snapshot.block_reason,
+                updated_at_epoch_secs: drain_gate_snapshot.updated_at_epoch_secs,
+            },
             dc_rtt: status
                 .dcs
                 .into_iter()
@@ -397,6 +488,81 @@ pub(super) async fn build_runtime_me_quality_data(shared: &ApiShared) -> Runtime
     }
 }
 
+fn build_runtime_me_teardown_data(shared: &ApiShared) -> RuntimeMeQualityTeardownData {
+    let attempts = MeWriterTeardownReason::ALL
+        .iter()
+        .copied()
+        .flat_map(|reason| {
+            MeWriterTeardownMode::ALL
+                .iter()
+                .copied()
+                .map(move |mode| RuntimeMeQualityTeardownAttemptData {
+                    reason: reason.as_str(),
+                    mode: mode.as_str(),
+                    total: shared.stats.get_me_writer_teardown_attempt_total(reason, mode),
+                })
+        })
+        .collect();
+
+    let success = MeWriterTeardownMode::ALL
+        .iter()
+        .copied()
+        .map(|mode| RuntimeMeQualityTeardownSuccessData {
+            mode: mode.as_str(),
+            total: shared.stats.get_me_writer_teardown_success_total(mode),
+        })
+        .collect();
+
+    let cleanup_side_effect_failures = MeWriterCleanupSideEffectStep::ALL
+        .iter()
+        .copied()
+        .map(|step| RuntimeMeQualityTeardownSideEffectData {
+            step: step.as_str(),
+            total: shared
+                .stats
+                .get_me_writer_cleanup_side_effect_failures_total(step),
+        })
+        .collect();
+
+    let duration = MeWriterTeardownMode::ALL
+        .iter()
+        .copied()
+        .map(|mode| {
+            let count = shared.stats.get_me_writer_teardown_duration_count(mode);
+            let mut buckets: Vec<RuntimeMeQualityTeardownDurationBucketData> = Stats::me_writer_teardown_duration_bucket_labels()
+                .iter()
+                .enumerate()
+                .map(|(bucket_idx, label)| RuntimeMeQualityTeardownDurationBucketData {
+                    le_seconds: label,
+                    total: shared
+                        .stats
+                        .get_me_writer_teardown_duration_bucket_total(mode, bucket_idx),
+                })
+                .collect();
+            buckets.push(RuntimeMeQualityTeardownDurationBucketData {
+                le_seconds: "+Inf",
+                total: count,
+            });
+            RuntimeMeQualityTeardownDurationData {
+                mode: mode.as_str(),
+                count,
+                sum_seconds: shared.stats.get_me_writer_teardown_duration_sum_seconds(mode),
+                buckets,
+            }
+        })
+        .collect();
+
+    RuntimeMeQualityTeardownData {
+        attempts,
+        success,
+        timeout_total: shared.stats.get_me_writer_teardown_timeout_total(),
+        escalation_total: shared.stats.get_me_writer_teardown_escalation_total(),
+        noop_total: shared.stats.get_me_writer_teardown_noop_total(),
+        cleanup_side_effect_failures,
+        duration,
+    }
+}
+
 pub(super) async fn build_runtime_upstream_quality_data(
     shared: &ApiShared,
 ) -> RuntimeUpstreamQualityData {
@@ -406,7 +572,9 @@ pub(super) async fn build_runtime_upstream_quality_data(
         connect_attempt_total: shared.stats.get_upstream_connect_attempt_total(),
         connect_success_total: shared.stats.get_upstream_connect_success_total(),
         connect_fail_total: shared.stats.get_upstream_connect_fail_total(),
-        connect_failfast_hard_error_total: shared.stats.get_upstream_connect_failfast_hard_error_total(),
+        connect_failfast_hard_error_total: shared
+            .stats
+            .get_upstream_connect_failfast_hard_error_total(),
     };
 
     let Some(snapshot) = shared.upstream_manager.try_api_snapshot() else {
@@ -446,6 +614,7 @@ pub(super) async fn build_runtime_upstream_quality_data(
             direct_total: snapshot.summary.direct_total,
             socks4_total: snapshot.summary.socks4_total,
             socks5_total: snapshot.summary.socks5_total,
+            shadowsocks_total: snapshot.summary.shadowsocks_total,
         }),
         upstreams: Some(
             snapshot
@@ -457,6 +626,7 @@ pub(super) async fn build_runtime_upstream_quality_data(
                         crate::transport::UpstreamRouteKind::Direct => "direct",
                         crate::transport::UpstreamRouteKind::Socks4 => "socks4",
                         crate::transport::UpstreamRouteKind::Socks5 => "socks5",
+                        crate::transport::UpstreamRouteKind::Shadowsocks => "shadowsocks",
                     },
                     address: upstream.address,
                     weight: upstream.weight,
@@ -476,7 +646,9 @@ pub(super) async fn build_runtime_upstream_quality_data(
                                 crate::transport::upstream::IpPreference::PreferV6 => "prefer_v6",
                                 crate::transport::upstream::IpPreference::PreferV4 => "prefer_v4",
                                 crate::transport::upstream::IpPreference::BothWork => "both_work",
-                                crate::transport::upstream::IpPreference::Unavailable => "unavailable",
+                                crate::transport::upstream::IpPreference::Unavailable => {
+                                    "unavailable"
+                                }
                             },
                         })
                         .collect(),
@@ -514,14 +686,18 @@ pub(super) async fn build_runtime_nat_stun_data(shared: &ApiShared) -> RuntimeNa
                 live_total: snapshot.live_servers.len(),
             },
             reflection: RuntimeNatStunReflectionBlockData {
-                v4: snapshot.reflection_v4.map(|entry| RuntimeNatStunReflectionData {
+                v4: snapshot
-                    addr: entry.addr.to_string(),
+                    .reflection_v4
-                    age_secs: entry.age_secs,
+                    .map(|entry| RuntimeNatStunReflectionData {
-                }),
+                        addr: entry.addr.to_string(),
-                v6: snapshot.reflection_v6.map(|entry| RuntimeNatStunReflectionData {
+                        age_secs: entry.age_secs,
-                    addr: entry.addr.to_string(),
+                    }),
-                    age_secs: entry.age_secs,
+                v6: snapshot
-                }),
+                    .reflection_v6
+                    .map(|entry| RuntimeNatStunReflectionData {
+                        addr: entry.addr.to_string(),
+                        age_secs: entry.age_secs,
+                    }),
             },
             stun_backoff_remaining_ms: snapshot.stun_backoff_remaining_ms,
         }),

src/api/runtime_selftest.rs

@@ -1,5 +1,5 @@
-use std::net::IpAddr;
 use std::collections::HashMap;
+use std::net::IpAddr;
 use std::sync::{Mutex, OnceLock};
 use std::time::{SystemTime, UNIX_EPOCH};
 
@@ -7,8 +7,8 @@ use serde::Serialize;
 
 use crate::config::{ProxyConfig, UpstreamType};
 use crate::network::probe::{detect_interface_ipv4, detect_interface_ipv6, is_bogon};
-use crate::transport::middle_proxy::{bnd_snapshot, timeskew_snapshot, upstream_bnd_snapshots};
 use crate::transport::UpstreamRouteKind;
+use crate::transport::middle_proxy::{bnd_snapshot, timeskew_snapshot, upstream_bnd_snapshots};
 
 use super::ApiShared;
 
@@ -262,8 +262,8 @@ fn update_kdf_ewma(now_epoch_secs: u64, total_errors: u64) -> f64 {
     let delta_errors = total_errors.saturating_sub(guard.last_total_errors);
     let instant_rate_per_min = (delta_errors as f64) * 60.0 / (dt_secs as f64);
     let alpha = 1.0 - f64::exp(-(dt_secs as f64) / KDF_EWMA_TAU_SECS);
-    guard.ewma_errors_per_min = guard.ewma_errors_per_min
+    guard.ewma_errors_per_min =
-        + alpha * (instant_rate_per_min - guard.ewma_errors_per_min);
+        guard.ewma_errors_per_min + alpha * (instant_rate_per_min - guard.ewma_errors_per_min);
     guard.last_epoch_secs = now_epoch_secs;
     guard.last_total_errors = total_errors;
     guard.ewma_errors_per_min
@@ -284,6 +284,7 @@ fn map_route_kind(value: UpstreamRouteKind) -> &'static str {
         UpstreamRouteKind::Direct => "direct",
         UpstreamRouteKind::Socks4 => "socks4",
         UpstreamRouteKind::Socks5 => "socks5",
+        UpstreamRouteKind::Shadowsocks => "shadowsocks",
     }
 }
 

src/api/runtime_stats.rs

@@ -1,7 +1,7 @@
 use std::time::{Duration, Instant, SystemTime, UNIX_EPOCH};
 
 use crate::config::ApiConfig;
-use crate::stats::Stats;
+use crate::stats::{MeWriterTeardownMode, Stats};
 use crate::transport::upstream::IpPreference;
 use crate::transport::UpstreamRouteKind;
 
@@ -106,6 +106,29 @@ pub(super) fn build_zero_all_data(stats: &Stats, configured_users: usize) -> Zer
             refill_failed_total: stats.get_me_refill_failed_total(),
             writer_restored_same_endpoint_total: stats.get_me_writer_restored_same_endpoint_total(),
             writer_restored_fallback_total: stats.get_me_writer_restored_fallback_total(),
+            teardown_attempt_total_normal: stats
+                .get_me_writer_teardown_attempt_total_by_mode(MeWriterTeardownMode::Normal),
+            teardown_attempt_total_hard_detach: stats
+                .get_me_writer_teardown_attempt_total_by_mode(MeWriterTeardownMode::HardDetach),
+            teardown_success_total_normal: stats
+                .get_me_writer_teardown_success_total(MeWriterTeardownMode::Normal),
+            teardown_success_total_hard_detach: stats
+                .get_me_writer_teardown_success_total(MeWriterTeardownMode::HardDetach),
+            teardown_timeout_total: stats.get_me_writer_teardown_timeout_total(),
+            teardown_escalation_total: stats.get_me_writer_teardown_escalation_total(),
+            teardown_noop_total: stats.get_me_writer_teardown_noop_total(),
+            teardown_cleanup_side_effect_failures_total: stats
+                .get_me_writer_cleanup_side_effect_failures_total_all(),
+            teardown_duration_count_total: stats
+                .get_me_writer_teardown_duration_count(MeWriterTeardownMode::Normal)
+                .saturating_add(
+                    stats.get_me_writer_teardown_duration_count(MeWriterTeardownMode::HardDetach),
+                ),
+            teardown_duration_sum_seconds_total: stats
+                .get_me_writer_teardown_duration_sum_seconds(MeWriterTeardownMode::Normal)
+                + stats.get_me_writer_teardown_duration_sum_seconds(
+                    MeWriterTeardownMode::HardDetach,
+                ),
         },
         desync: ZeroDesyncData {
             secure_padding_invalid_total: stats.get_secure_padding_invalid(),
@@ -138,7 +161,8 @@ fn build_zero_upstream_data(stats: &Stats) -> ZeroUpstreamData {
             .get_upstream_connect_duration_success_bucket_501_1000ms(),
         connect_duration_success_bucket_gt_1000ms: stats
             .get_upstream_connect_duration_success_bucket_gt_1000ms(),
-        connect_duration_fail_bucket_le_100ms: stats.get_upstream_connect_duration_fail_bucket_le_100ms(),
+        connect_duration_fail_bucket_le_100ms: stats
+            .get_upstream_connect_duration_fail_bucket_le_100ms(),
         connect_duration_fail_bucket_101_500ms: stats
             .get_upstream_connect_duration_fail_bucket_101_500ms(),
         connect_duration_fail_bucket_501_1000ms: stats
@@ -180,6 +204,7 @@ pub(super) fn build_upstreams_data(shared: &ApiShared, api_cfg: &ApiConfig) -> U
         direct_total: snapshot.summary.direct_total,
         socks4_total: snapshot.summary.socks4_total,
         socks5_total: snapshot.summary.socks5_total,
+        shadowsocks_total: snapshot.summary.shadowsocks_total,
     };
     let upstreams = snapshot
         .upstreams
@@ -395,8 +420,7 @@ async fn get_minimal_payload_cached(
         adaptive_floor_min_writers_multi_endpoint: runtime
             .adaptive_floor_min_writers_multi_endpoint,
         adaptive_floor_recover_grace_secs: runtime.adaptive_floor_recover_grace_secs,
-        adaptive_floor_writers_per_core_total: runtime
+        adaptive_floor_writers_per_core_total: runtime.adaptive_floor_writers_per_core_total,
-            .adaptive_floor_writers_per_core_total,
         adaptive_floor_cpu_cores_override: runtime.adaptive_floor_cpu_cores_override,
         adaptive_floor_max_extra_writers_single_per_core: runtime
             .adaptive_floor_max_extra_writers_single_per_core,
@@ -404,12 +428,9 @@ async fn get_minimal_payload_cached(
             .adaptive_floor_max_extra_writers_multi_per_core,
         adaptive_floor_max_active_writers_per_core: runtime
             .adaptive_floor_max_active_writers_per_core,
-        adaptive_floor_max_warm_writers_per_core: runtime
+        adaptive_floor_max_warm_writers_per_core: runtime.adaptive_floor_max_warm_writers_per_core,
-            .adaptive_floor_max_warm_writers_per_core,
+        adaptive_floor_max_active_writers_global: runtime.adaptive_floor_max_active_writers_global,
-        adaptive_floor_max_active_writers_global: runtime
+        adaptive_floor_max_warm_writers_global: runtime.adaptive_floor_max_warm_writers_global,
-            .adaptive_floor_max_active_writers_global,
-        adaptive_floor_max_warm_writers_global: runtime
-            .adaptive_floor_max_warm_writers_global,
         adaptive_floor_cpu_cores_detected: runtime.adaptive_floor_cpu_cores_detected,
         adaptive_floor_cpu_cores_effective: runtime.adaptive_floor_cpu_cores_effective,
         adaptive_floor_global_cap_raw: runtime.adaptive_floor_global_cap_raw,
@@ -528,6 +549,7 @@ fn map_route_kind(value: UpstreamRouteKind) -> &'static str {
         UpstreamRouteKind::Direct => "direct",
         UpstreamRouteKind::Socks4 => "socks4",
         UpstreamRouteKind::Socks5 => "socks5",
+        UpstreamRouteKind::Shadowsocks => "shadowsocks",
     }
 }
 

src/cli.rs

@@ -199,8 +199,14 @@ update_every = 43200
 hardswap = false
 me_pool_drain_ttl_secs = 90
 me_instadrain = false
+me_pool_drain_threshold = 32
+me_pool_drain_soft_evict_grace_secs = 10
+me_pool_drain_soft_evict_per_writer = 2
+me_pool_drain_soft_evict_budget_per_core = 16
+me_pool_drain_soft_evict_cooldown_ms = 1000
+me_bind_stale_mode = "never"
 me_pool_min_fresh_ratio = 0.8
-me_reinit_drain_timeout_secs = 120
+me_reinit_drain_timeout_secs = 90
 
 [network]
 ipv4 = true
@@ -262,7 +268,7 @@ fn generate_systemd_unit(exe_path: &Path, config_path: &Path) -> String {
     format!(
 r#"[Unit]
 Description=Telemt MTProxy
-Documentation=https://github.com/nicepkg/telemt
+Documentation=https://github.com/telemt/telemt
 After=network-online.target
 Wants=network-online.target
 

src/config/defaults.rs

@@ -40,10 +40,10 @@ const DEFAULT_ME_ROUTE_HYBRID_MAX_WAIT_MS: u64 = 3000;
 const DEFAULT_ME_ROUTE_BLOCKING_SEND_TIMEOUT_MS: u64 = 250;
 const DEFAULT_ME_C2ME_SEND_TIMEOUT_MS: u64 = 4000;
 const DEFAULT_ME_POOL_DRAIN_SOFT_EVICT_ENABLED: bool = true;
-const DEFAULT_ME_POOL_DRAIN_SOFT_EVICT_GRACE_SECS: u64 = 30;
+const DEFAULT_ME_POOL_DRAIN_SOFT_EVICT_GRACE_SECS: u64 = 10;
-const DEFAULT_ME_POOL_DRAIN_SOFT_EVICT_PER_WRITER: u8 = 1;
+const DEFAULT_ME_POOL_DRAIN_SOFT_EVICT_PER_WRITER: u8 = 2;
-const DEFAULT_ME_POOL_DRAIN_SOFT_EVICT_BUDGET_PER_CORE: u16 = 8;
+const DEFAULT_ME_POOL_DRAIN_SOFT_EVICT_BUDGET_PER_CORE: u16 = 16;
-const DEFAULT_ME_POOL_DRAIN_SOFT_EVICT_COOLDOWN_MS: u64 = 5000;
+const DEFAULT_ME_POOL_DRAIN_SOFT_EVICT_COOLDOWN_MS: u64 = 1000;
 const DEFAULT_USER_MAX_UNIQUE_IPS_WINDOW_SECS: u64 = 30;
 const DEFAULT_ACCEPT_PERMIT_TIMEOUT_MS: u64 = 250;
 const DEFAULT_UPSTREAM_CONNECT_RETRY_ATTEMPTS: u32 = 2;
@@ -65,6 +65,10 @@ pub(crate) fn default_tls_domain() -> String {
     "petrovich.ru".to_string()
 }
 
+pub(crate) fn default_tls_fetch_scope() -> String {
+    String::new()
+}
+
 pub(crate) fn default_mask_port() -> u16 {
     443
 }
@@ -606,7 +610,7 @@ pub(crate) fn default_proxy_secret_len_max() -> usize {
 }
 
 pub(crate) fn default_me_reinit_drain_timeout_secs() -> u64 {
-    120
+    90
 }
 
 pub(crate) fn default_me_pool_drain_ttl_secs() -> u64 {
@@ -618,7 +622,7 @@ pub(crate) fn default_me_instadrain() -> bool {
 }
 
 pub(crate) fn default_me_pool_drain_threshold() -> u64 {
-    128
+    32
 }
 
 pub(crate) fn default_me_pool_drain_soft_evict_enabled() -> bool {

src/config/hot_reload.rs

@@ -623,6 +623,7 @@ fn warn_non_hot_changes(old: &ProxyConfig, new: &ProxyConfig, non_hot_changed: b
     }
     if old.censorship.tls_domain != new.censorship.tls_domain
         || old.censorship.tls_domains != new.censorship.tls_domains
+        || old.censorship.tls_fetch_scope != new.censorship.tls_fetch_scope
         || old.censorship.mask != new.censorship.mask
         || old.censorship.mask_host != new.censorship.mask_host
         || old.censorship.mask_port != new.censorship.mask_port

src/config/load.rs

@@ -6,8 +6,9 @@ use std::net::{IpAddr, SocketAddr};
 use std::path::{Path, PathBuf};
 
 use rand::Rng;
+use serde::{Deserialize, Serialize};
+use shadowsocks::config::ServerConfig as ShadowsocksServerConfig;
 use tracing::warn;
-use serde::{Serialize, Deserialize};
 
 use crate::error::{ProxyError, Result};
 
@@ -122,13 +123,37 @@ fn sanitize_ad_tag(ad_tag: &mut Option<String>) {
     };
 
     if !is_valid_ad_tag(tag) {
-        warn!(
+        warn!("Invalid general.ad_tag value, expected exactly 32 hex chars; ad_tag is disabled");
-            "Invalid general.ad_tag value, expected exactly 32 hex chars; ad_tag is disabled"
-        );
         *ad_tag = None;
     }
 }
 
+fn validate_upstreams(config: &ProxyConfig) -> Result<()> {
+    let has_enabled_shadowsocks = config.upstreams.iter().any(|upstream| {
+        upstream.enabled && matches!(upstream.upstream_type, UpstreamType::Shadowsocks { .. })
+    });
+
+    if has_enabled_shadowsocks && config.general.use_middle_proxy {
+        return Err(ProxyError::Config(
+            "shadowsocks upstreams require general.use_middle_proxy = false".to_string(),
+        ));
+    }
+
+    for upstream in &config.upstreams {
+        if let UpstreamType::Shadowsocks { url, .. } = &upstream.upstream_type {
+            let parsed = ShadowsocksServerConfig::from_url(url)
+                .map_err(|error| ProxyError::Config(format!("invalid shadowsocks url: {error}")))?;
+            if parsed.plugin().is_some() {
+                return Err(ProxyError::Config(
+                    "shadowsocks plugins are not supported".to_string(),
+                ));
+            }
+        }
+    }
+
+    Ok(())
+}
+
 // ============= Main Config =============
 
 #[derive(Debug, Clone, Serialize, Deserialize, Default)]
@@ -180,7 +205,8 @@ impl ProxyConfig {
 
     pub(crate) fn load_with_metadata<P: AsRef<Path>>(path: P) -> Result<LoadedConfig> {
         let path = path.as_ref();
-        let content = std::fs::read_to_string(path).map_err(|e| ProxyError::Config(e.to_string()))?;
+        let content =
+            std::fs::read_to_string(path).map_err(|e| ProxyError::Config(e.to_string()))?;
         let base_dir = path.parent().unwrap_or(Path::new("."));
         let mut source_files = BTreeSet::new();
         source_files.insert(normalize_config_path(path));
@@ -207,15 +233,17 @@ impl ProxyConfig {
             .map(|table| table.contains_key("stun_servers"))
             .unwrap_or(false);
 
-        let mut config: ProxyConfig =
+        let mut config: ProxyConfig = parsed_toml
-            parsed_toml.try_into().map_err(|e| ProxyError::Config(e.to_string()))?;
+            .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_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();
@@ -225,7 +253,9 @@ impl ProxyConfig {
             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");
+                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.
@@ -240,7 +270,9 @@ impl ProxyConfig {
             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");
+                warn!(
+                    "general.middle_proxy_nat_stun and general.middle_proxy_nat_stun_servers are deprecated; use network.stun_servers"
+                );
             }
         }
 
@@ -378,13 +410,15 @@ impl ProxyConfig {
 
         if !(4096..=1024 * 1024).contains(&config.general.direct_relay_copy_buf_c2s_bytes) {
             return Err(ProxyError::Config(
-                "general.direct_relay_copy_buf_c2s_bytes must be within [4096, 1048576]".to_string(),
+                "general.direct_relay_copy_buf_c2s_bytes must be within [4096, 1048576]"
+                    .to_string(),
             ));
         }
 
         if !(8192..=2 * 1024 * 1024).contains(&config.general.direct_relay_copy_buf_s2c_bytes) {
             return Err(ProxyError::Config(
-                "general.direct_relay_copy_buf_s2c_bytes must be within [8192, 2097152]".to_string(),
+                "general.direct_relay_copy_buf_s2c_bytes must be within [8192, 2097152]"
+                    .to_string(),
             ));
         }
 
@@ -633,7 +667,8 @@ impl ProxyConfig {
 
         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(),
+                "general.me_route_backpressure_high_watermark_pct must be within [1, 100]"
+                    .to_string(),
             ));
         }
 
@@ -779,6 +814,9 @@ impl ProxyConfig {
             config.censorship.mask_host = Some(config.censorship.tls_domain.clone());
         }
 
+        // Normalize optional TLS fetch scope: whitespace-only values disable scoped routing.
+        config.censorship.tls_fetch_scope = config.censorship.tls_fetch_scope.trim().to_string();
+
         // Merge primary + extra TLS domains, deduplicate (primary always first).
         if !config.censorship.tls_domains.is_empty() {
             let mut all = Vec::with_capacity(1 + config.censorship.tls_domains.len());
@@ -813,11 +851,15 @@ impl ProxyConfig {
         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");
+            warn!(
+                "IPv6 with Middle Proxy is experimental and may cause KDF address mismatch; consider disabling IPv6 or ME"
+            );
         }
 
         // Random fake_cert_len only when default is in use.
-        if !config.censorship.tls_emulation && config.censorship.fake_cert_len == default_fake_cert_len() {
+        if !config.censorship.tls_emulation
+            && config.censorship.fake_cert_len == default_fake_cert_len()
+        {
             config.censorship.fake_cert_len = rand::rng().gen_range(1024..4096);
         }
 
@@ -827,8 +869,7 @@ impl ProxyConfig {
         let listen_tcp = config.server.listen_tcp.unwrap_or_else(|| {
             if config.server.listen_unix_sock.is_some() {
                 // Unix socket present: TCP only if user explicitly set addresses or listeners.
-                config.server.listen_addr_ipv4.is_some()
+                config.server.listen_addr_ipv4.is_some() || !config.server.listeners.is_empty()
-                    || !config.server.listeners.is_empty()
             } else {
                 true
             }
@@ -836,7 +877,9 @@ impl ProxyConfig {
 
         // Migration: Populate listeners if empty (skip when listen_tcp = false).
         if config.server.listeners.is_empty() && listen_tcp {
-            let ipv4_str = config.server.listen_addr_ipv4
+            let ipv4_str = config
+                .server
+                .listen_addr_ipv4
                 .as_deref()
                 .unwrap_or("0.0.0.0");
             if let Ok(ipv4) = ipv4_str.parse::<IpAddr>() {
@@ -878,7 +921,10 @@ impl ProxyConfig {
         // Migration: Populate upstreams if empty (Default Direct).
         if config.upstreams.is_empty() {
             config.upstreams.push(UpstreamConfig {
-                upstream_type: UpstreamType::Direct { interface: None, bind_addresses: None },
+                upstream_type: UpstreamType::Direct {
+                    interface: None,
+                    bind_addresses: None,
+                },
                 weight: 1,
                 enabled: true,
                 scopes: String::new(),
@@ -892,6 +938,8 @@ impl ProxyConfig {
             .entry("203".to_string())
             .or_insert_with(|| vec!["91.105.192.100:443".to_string()]);
 
+        validate_upstreams(&config)?;
+
         Ok(LoadedConfig {
             config,
             source_files: source_files.into_iter().collect(),
@@ -938,6 +986,9 @@ impl ProxyConfig {
 mod tests {
     use super::*;
 
+    const TEST_SHADOWSOCKS_URL: &str =
+        "ss://2022-blake3-aes-256-gcm:MDEyMzQ1Njc4OTAxMjM0NTY3ODkwMTIzNDU2Nzg5MDE=@127.0.0.1:8388";
+
     #[test]
     fn serde_defaults_remain_unchanged_for_present_sections() {
         let toml = r#"
@@ -967,10 +1018,7 @@ mod tests {
             cfg.general.me_init_retry_attempts,
             default_me_init_retry_attempts()
         );
-        assert_eq!(
+        assert_eq!(cfg.general.me2dc_fallback, default_me2dc_fallback());
-            cfg.general.me2dc_fallback,
-            default_me2dc_fallback()
-        );
         assert_eq!(
             cfg.general.proxy_config_v4_cache_path,
             default_proxy_config_v4_cache_path()
@@ -1279,11 +1327,12 @@ mod tests {
         let path = dir.join("telemt_dc_override_test.toml");
         std::fs::write(&path, toml).unwrap();
         let cfg = ProxyConfig::load(&path).unwrap();
-        assert!(cfg
+        assert!(
-            .dc_overrides
+            cfg.dc_overrides
-            .get("203")
+                .get("203")
-            .map(|v| v.contains(&"91.105.192.100:443".to_string()))
+                .map(|v| v.contains(&"91.105.192.100:443".to_string()))
-            .unwrap_or(false));
+                .unwrap_or(false)
+        );
         let _ = std::fs::remove_file(path);
     }
 
@@ -1470,11 +1519,9 @@ mod tests {
         let path = dir.join("telemt_me_adaptive_floor_min_writers_out_of_range_test.toml");
         std::fs::write(&path, toml).unwrap();
         let err = ProxyConfig::load(&path).unwrap_err().to_string();
-        assert!(
+        assert!(err.contains(
-            err.contains(
+            "general.me_adaptive_floor_min_writers_single_endpoint must be within [1, 32]"
-                "general.me_adaptive_floor_min_writers_single_endpoint must be within [1, 32]"
+        ));
-            )
-        );
         let _ = std::fs::remove_file(path);
     }
 
@@ -2037,6 +2084,45 @@ mod tests {
         let _ = std::fs::remove_file(path);
     }
 
+    #[test]
+    fn force_close_default_matches_drain_ttl() {
+        let toml = r#"
+            [censorship]
+            tls_domain = "example.com"
+
+            [access.users]
+            user = "00000000000000000000000000000000"
+        "#;
+        let dir = std::env::temp_dir();
+        let path = dir.join("telemt_force_close_default_test.toml");
+        std::fs::write(&path, toml).unwrap();
+        let cfg = ProxyConfig::load(&path).unwrap();
+        assert_eq!(cfg.general.me_reinit_drain_timeout_secs, 90);
+        assert_eq!(cfg.general.effective_me_pool_force_close_secs(), 90);
+        let _ = std::fs::remove_file(path);
+    }
+
+    #[test]
+    fn force_close_zero_uses_runtime_safety_fallback() {
+        let toml = r#"
+            [general]
+            me_reinit_drain_timeout_secs = 0
+
+            [censorship]
+            tls_domain = "example.com"
+
+            [access.users]
+            user = "00000000000000000000000000000000"
+        "#;
+        let dir = std::env::temp_dir();
+        let path = dir.join("telemt_force_close_zero_fallback_test.toml");
+        std::fs::write(&path, toml).unwrap();
+        let cfg = ProxyConfig::load(&path).unwrap();
+        assert_eq!(cfg.general.me_reinit_drain_timeout_secs, 0);
+        assert_eq!(cfg.general.effective_me_pool_force_close_secs(), 300);
+        let _ = std::fs::remove_file(path);
+    }
+
     #[test]
     fn force_close_bumped_when_below_drain_ttl() {
         let toml = r#"
@@ -2058,6 +2144,59 @@ mod tests {
         let _ = std::fs::remove_file(path);
     }
 
+    #[test]
+    fn tls_fetch_scope_default_is_empty() {
+        let toml = r#"
+            [censorship]
+            tls_domain = "example.com"
+
+            [access.users]
+            user = "00000000000000000000000000000000"
+        "#;
+        let dir = std::env::temp_dir();
+        let path = dir.join("telemt_tls_fetch_scope_default_test.toml");
+        std::fs::write(&path, toml).unwrap();
+        let cfg = ProxyConfig::load(&path).unwrap();
+        assert!(cfg.censorship.tls_fetch_scope.is_empty());
+        let _ = std::fs::remove_file(path);
+    }
+
+    #[test]
+    fn tls_fetch_scope_is_trimmed_during_load() {
+        let toml = r#"
+            [censorship]
+            tls_domain = "example.com"
+            tls_fetch_scope = "  me  "
+
+            [access.users]
+            user = "00000000000000000000000000000000"
+        "#;
+        let dir = std::env::temp_dir();
+        let path = dir.join("telemt_tls_fetch_scope_trim_test.toml");
+        std::fs::write(&path, toml).unwrap();
+        let cfg = ProxyConfig::load(&path).unwrap();
+        assert_eq!(cfg.censorship.tls_fetch_scope, "me");
+        let _ = std::fs::remove_file(path);
+    }
+
+    #[test]
+    fn tls_fetch_scope_whitespace_becomes_empty() {
+        let toml = r#"
+            [censorship]
+            tls_domain = "example.com"
+            tls_fetch_scope = "   "
+
+            [access.users]
+            user = "00000000000000000000000000000000"
+        "#;
+        let dir = std::env::temp_dir();
+        let path = dir.join("telemt_tls_fetch_scope_blank_test.toml");
+        std::fs::write(&path, toml).unwrap();
+        let cfg = ProxyConfig::load(&path).unwrap();
+        assert!(cfg.censorship.tls_fetch_scope.is_empty());
+        let _ = std::fs::remove_file(path);
+    }
+
     #[test]
     fn invalid_ad_tag_is_disabled_during_load() {
         let toml = r#"
@@ -2101,6 +2240,124 @@ mod tests {
         let _ = std::fs::remove_file(path);
     }
 
+    #[test]
+    fn shadowsocks_upstream_url_loads_successfully() {
+        let toml = format!(
+            r#"
+            [general]
+            use_middle_proxy = false
+
+            [censorship]
+            tls_domain = "example.com"
+
+            [access.users]
+            user = "00000000000000000000000000000000"
+
+            [[upstreams]]
+            type = "shadowsocks"
+            url = "{url}"
+            interface = "127.0.0.2"
+            "#,
+            url = TEST_SHADOWSOCKS_URL,
+        );
+        let dir = std::env::temp_dir();
+        let path = dir.join("telemt_shadowsocks_valid_test.toml");
+        std::fs::write(&path, toml).unwrap();
+        let cfg = ProxyConfig::load(&path).unwrap();
+
+        assert!(matches!(
+            &cfg.upstreams[0].upstream_type,
+            UpstreamType::Shadowsocks { url, interface }
+                if url == TEST_SHADOWSOCKS_URL && interface.as_deref() == Some("127.0.0.2")
+        ));
+
+        let _ = std::fs::remove_file(path);
+    }
+
+    #[test]
+    fn shadowsocks_requires_direct_mode() {
+        let toml = format!(
+            r#"
+            [general]
+            use_middle_proxy = true
+
+            [censorship]
+            tls_domain = "example.com"
+
+            [access.users]
+            user = "00000000000000000000000000000000"
+
+            [[upstreams]]
+            type = "shadowsocks"
+            url = "{url}"
+            "#,
+            url = TEST_SHADOWSOCKS_URL,
+        );
+        let dir = std::env::temp_dir();
+        let path = dir.join("telemt_shadowsocks_me_reject_test.toml");
+        std::fs::write(&path, toml).unwrap();
+        let err = ProxyConfig::load(&path).unwrap_err().to_string();
+
+        assert!(err.contains("shadowsocks upstreams require general.use_middle_proxy = false"));
+
+        let _ = std::fs::remove_file(path);
+    }
+
+    #[test]
+    fn invalid_shadowsocks_url_is_rejected() {
+        let toml = r#"
+            [general]
+            use_middle_proxy = false
+
+            [censorship]
+            tls_domain = "example.com"
+
+            [access.users]
+            user = "00000000000000000000000000000000"
+
+            [[upstreams]]
+            type = "shadowsocks"
+            url = "not-a-valid-ss-url"
+        "#;
+        let dir = std::env::temp_dir();
+        let path = dir.join("telemt_shadowsocks_invalid_url_test.toml");
+        std::fs::write(&path, toml).unwrap();
+        let err = ProxyConfig::load(&path).unwrap_err().to_string();
+
+        assert!(err.contains("invalid shadowsocks url"));
+
+        let _ = std::fs::remove_file(path);
+    }
+
+    #[test]
+    fn shadowsocks_plugins_are_rejected() {
+        let toml = format!(
+            r#"
+            [general]
+            use_middle_proxy = false
+
+            [censorship]
+            tls_domain = "example.com"
+
+            [access.users]
+            user = "00000000000000000000000000000000"
+
+            [[upstreams]]
+            type = "shadowsocks"
+            url = "{url}?plugin=obfs-local%3Bobfs%3Dhttp"
+            "#,
+            url = TEST_SHADOWSOCKS_URL,
+        );
+        let dir = std::env::temp_dir();
+        let path = dir.join("telemt_shadowsocks_plugin_reject_test.toml");
+        std::fs::write(&path, toml).unwrap();
+        let err = ProxyConfig::load(&path).unwrap_err().to_string();
+
+        assert!(err.contains("shadowsocks plugins are not supported"));
+
+        let _ = std::fs::remove_file(path);
+    }
+
     #[test]
     fn invalid_user_ad_tag_reports_access_user_ad_tags_key() {
         let toml = r#"

src/config/types.rs

@@ -135,8 +135,8 @@ impl MeSocksKdfPolicy {
 #[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize, Default)]
 #[serde(rename_all = "lowercase")]
 pub enum MeBindStaleMode {
-    Never,
     #[default]
+    Never,
     Ttl,
     Always,
 }
@@ -855,7 +855,7 @@ pub struct GeneralConfig {
     pub me_pool_min_fresh_ratio: f32,
 
     /// Drain timeout in seconds for stale ME writers after endpoint map changes.
-    /// Set to 0 to keep stale writers draining indefinitely (no force-close).
+    /// Set to 0 to use the runtime safety fallback timeout.
     #[serde(default = "default_me_reinit_drain_timeout_secs")]
     pub me_reinit_drain_timeout_secs: u64,
 
@@ -955,24 +955,38 @@ impl Default for GeneralConfig {
             me_reconnect_backoff_cap_ms: default_reconnect_backoff_cap_ms(),
             me_reconnect_fast_retry_count: default_me_reconnect_fast_retry_count(),
             me_single_endpoint_shadow_writers: default_me_single_endpoint_shadow_writers(),
-            me_single_endpoint_outage_mode_enabled: default_me_single_endpoint_outage_mode_enabled(),
+            me_single_endpoint_outage_mode_enabled: default_me_single_endpoint_outage_mode_enabled(
-            me_single_endpoint_outage_disable_quarantine: default_me_single_endpoint_outage_disable_quarantine(),
+            ),
-            me_single_endpoint_outage_backoff_min_ms: default_me_single_endpoint_outage_backoff_min_ms(),
+            me_single_endpoint_outage_disable_quarantine:
-            me_single_endpoint_outage_backoff_max_ms: default_me_single_endpoint_outage_backoff_max_ms(),
+                default_me_single_endpoint_outage_disable_quarantine(),
-            me_single_endpoint_shadow_rotate_every_secs: default_me_single_endpoint_shadow_rotate_every_secs(),
+            me_single_endpoint_outage_backoff_min_ms:
+                default_me_single_endpoint_outage_backoff_min_ms(),
+            me_single_endpoint_outage_backoff_max_ms:
+                default_me_single_endpoint_outage_backoff_max_ms(),
+            me_single_endpoint_shadow_rotate_every_secs:
+                default_me_single_endpoint_shadow_rotate_every_secs(),
             me_floor_mode: MeFloorMode::default(),
             me_adaptive_floor_idle_secs: default_me_adaptive_floor_idle_secs(),
-            me_adaptive_floor_min_writers_single_endpoint: default_me_adaptive_floor_min_writers_single_endpoint(),
+            me_adaptive_floor_min_writers_single_endpoint:
-            me_adaptive_floor_min_writers_multi_endpoint: default_me_adaptive_floor_min_writers_multi_endpoint(),
+                default_me_adaptive_floor_min_writers_single_endpoint(),
+            me_adaptive_floor_min_writers_multi_endpoint:
+                default_me_adaptive_floor_min_writers_multi_endpoint(),
             me_adaptive_floor_recover_grace_secs: default_me_adaptive_floor_recover_grace_secs(),
-            me_adaptive_floor_writers_per_core_total: default_me_adaptive_floor_writers_per_core_total(),
+            me_adaptive_floor_writers_per_core_total:
+                default_me_adaptive_floor_writers_per_core_total(),
             me_adaptive_floor_cpu_cores_override: default_me_adaptive_floor_cpu_cores_override(),
-            me_adaptive_floor_max_extra_writers_single_per_core: default_me_adaptive_floor_max_extra_writers_single_per_core(),
+            me_adaptive_floor_max_extra_writers_single_per_core:
-            me_adaptive_floor_max_extra_writers_multi_per_core: default_me_adaptive_floor_max_extra_writers_multi_per_core(),
+                default_me_adaptive_floor_max_extra_writers_single_per_core(),
-            me_adaptive_floor_max_active_writers_per_core: default_me_adaptive_floor_max_active_writers_per_core(),
+            me_adaptive_floor_max_extra_writers_multi_per_core:
-            me_adaptive_floor_max_warm_writers_per_core: default_me_adaptive_floor_max_warm_writers_per_core(),
+                default_me_adaptive_floor_max_extra_writers_multi_per_core(),
-            me_adaptive_floor_max_active_writers_global: default_me_adaptive_floor_max_active_writers_global(),
+            me_adaptive_floor_max_active_writers_per_core:
-            me_adaptive_floor_max_warm_writers_global: default_me_adaptive_floor_max_warm_writers_global(),
+                default_me_adaptive_floor_max_active_writers_per_core(),
+            me_adaptive_floor_max_warm_writers_per_core:
+                default_me_adaptive_floor_max_warm_writers_per_core(),
+            me_adaptive_floor_max_active_writers_global:
+                default_me_adaptive_floor_max_active_writers_global(),
+            me_adaptive_floor_max_warm_writers_global:
+                default_me_adaptive_floor_max_warm_writers_global(),
             upstream_connect_retry_attempts: default_upstream_connect_retry_attempts(),
             upstream_connect_retry_backoff_ms: default_upstream_connect_retry_backoff_ms(),
             upstream_connect_budget_ms: default_upstream_connect_budget_ms(),
@@ -987,7 +1001,8 @@ impl Default for GeneralConfig {
             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(),
+            me_route_backpressure_high_watermark_pct:
+                default_me_route_backpressure_high_watermark_pct(),
             me_health_interval_ms_unhealthy: default_me_health_interval_ms_unhealthy(),
             me_health_interval_ms_healthy: default_me_health_interval_ms_healthy(),
             me_admission_poll_ms: default_me_admission_poll_ms(),
@@ -1013,7 +1028,8 @@ impl Default for GeneralConfig {
             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_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(),
             me_snapshot_require_http_2xx: default_me_snapshot_require_http_2xx(),
@@ -1057,8 +1073,10 @@ impl GeneralConfig {
     /// Resolve the active updater interval for ME infrastructure refresh tasks.
     /// `update_every` has priority, otherwise legacy proxy_*_auto_reload_secs are used.
     pub fn effective_update_every_secs(&self) -> u64 {
-        self.update_every
+        self.update_every.unwrap_or_else(|| {
-            .unwrap_or_else(|| self.proxy_secret_auto_reload_secs.min(self.proxy_config_auto_reload_secs))
+            self.proxy_secret_auto_reload_secs
+                .min(self.proxy_config_auto_reload_secs)
+        })
     }
 
     /// Resolve periodic zero-downtime reinit interval for ME writers.
@@ -1068,8 +1086,13 @@ impl GeneralConfig {
 
     /// Resolve force-close timeout for stale writers.
     /// `me_reinit_drain_timeout_secs` remains backward-compatible alias.
+    /// A configured `0` uses the runtime safety fallback (300s).
     pub fn effective_me_pool_force_close_secs(&self) -> u64 {
-        self.me_reinit_drain_timeout_secs
+        if self.me_reinit_drain_timeout_secs == 0 {
+            300
+        } else {
+            self.me_reinit_drain_timeout_secs
+        }
     }
 }
 
@@ -1303,6 +1326,11 @@ pub struct AntiCensorshipConfig {
     #[serde(default)]
     pub tls_domains: Vec<String>,
 
+    /// Upstream scope used for TLS front metadata fetches.
+    /// Empty value keeps default upstream routing behavior.
+    #[serde(default = "default_tls_fetch_scope")]
+    pub tls_fetch_scope: String,
+
     #[serde(default = "default_true")]
     pub mask: bool,
 
@@ -1360,6 +1388,7 @@ impl Default for AntiCensorshipConfig {
         Self {
             tls_domain: default_tls_domain(),
             tls_domains: Vec::new(),
+            tls_fetch_scope: default_tls_fetch_scope(),
             mask: default_true(),
             mask_host: None,
             mask_port: default_mask_port(),
@@ -1465,6 +1494,11 @@ pub enum UpstreamType {
         #[serde(default)]
         password: Option<String>,
     },
+    Shadowsocks {
+        url: String,
+        #[serde(default)]
+        interface: Option<String>,
+    },
 }
 
 #[derive(Debug, Clone, Serialize, Deserialize)]
@@ -1545,7 +1579,10 @@ impl ShowLink {
 }
 
 impl Serialize for ShowLink {
-    fn serialize<S: serde::Serializer>(&self, serializer: S) -> std::result::Result<S::Ok, S::Error> {
+    fn serialize<S: serde::Serializer>(
+        &self,
+        serializer: S,
+    ) -> std::result::Result<S::Ok, S::Error> {
         match self {
             ShowLink::None => Vec::<String>::new().serialize(serializer),
             ShowLink::All => serializer.serialize_str("*"),
@@ -1555,7 +1592,9 @@ impl Serialize for ShowLink {
 }
 
 impl<'de> Deserialize<'de> for ShowLink {
-    fn deserialize<D: serde::Deserializer<'de>>(deserializer: D) -> std::result::Result<Self, D::Error> {
+    fn deserialize<D: serde::Deserializer<'de>>(
+        deserializer: D,
+    ) -> std::result::Result<Self, D::Error> {
         use serde::de;
 
         struct ShowLinkVisitor;
@@ -1571,14 +1610,14 @@ impl<'de> Deserialize<'de> for ShowLink {
                 if v == "*" {
                     Ok(ShowLink::All)
                 } else {
-                    Err(de::Error::invalid_value(
+                    Err(de::Error::invalid_value(de::Unexpected::Str(v), &r#""*""#))
-                        de::Unexpected::Str(v),
-                        &r#""*""#,
-                    ))
                 }
             }
 
-            fn visit_seq<A: de::SeqAccess<'de>>(self, mut seq: A) -> std::result::Result<ShowLink, A::Error> {
+            fn visit_seq<A: de::SeqAccess<'de>>(
+                self,
+                mut seq: A,
+            ) -> std::result::Result<ShowLink, A::Error> {
                 let mut names = Vec::new();
                 while let Some(name) = seq.next_element::<String>()? {
                     names.push(name);

src/maestro/me_startup.rs

@@ -332,25 +332,76 @@ pub(crate) async fn initialize_me_pool(
                                             "Middle-End pool initialized successfully"
                                         );
 
-                                        let pool_health = pool_bg.clone();
+                                            // ── Supervised background tasks ──────────────────
-                                        let rng_health = rng_bg.clone();
+                                            // Each task runs inside a nested tokio::spawn so
-                                        let min_conns = pool_size;
+                                            // that a panic is caught via JoinHandle and the
-                                        tokio::spawn(async move {
+                                            // outer loop restarts the task automatically.
-                                            crate::transport::middle_proxy::me_health_monitor(
+                                            let pool_health = pool_bg.clone();
-                                                pool_health,
+                                            let rng_health = rng_bg.clone();
-                                                rng_health,
+                                            let min_conns = pool_size;
-                                                min_conns,
+                                            tokio::spawn(async move {
-                                            )
+                                                loop {
-                                            .await;
+                                                    let p = pool_health.clone();
-                                        });
+                                                    let r = rng_health.clone();
-                                        let pool_drain_enforcer = pool_bg.clone();
+                                                    let res = tokio::spawn(async move {
-                                        tokio::spawn(async move {
+                                                        crate::transport::middle_proxy::me_health_monitor(
-                                            crate::transport::middle_proxy::me_drain_timeout_enforcer(
+                                                            p, r, min_conns,
-                                                pool_drain_enforcer,
+                                                        )
-                                            )
+                                                        .await;
-                                            .await;
+                                                    })
-                                        });
+                                                    .await;
-                                        break;
+                                                    match res {
+                                                        Ok(()) => warn!("me_health_monitor exited unexpectedly, restarting"),
+                                                        Err(e) => {
+                                                            error!(error = %e, "me_health_monitor panicked, restarting in 1s");
+                                                            tokio::time::sleep(Duration::from_secs(1)).await;
+                                                        }
+                                                    }
+                                                }
+                                            });
+                                            let pool_drain_enforcer = pool_bg.clone();
+                                            tokio::spawn(async move {
+                                                loop {
+                                                    let p = pool_drain_enforcer.clone();
+                                                    let res = tokio::spawn(async move {
+                                                        crate::transport::middle_proxy::me_drain_timeout_enforcer(p).await;
+                                                    })
+                                                    .await;
+                                                    match res {
+                                                        Ok(()) => warn!("me_drain_timeout_enforcer exited unexpectedly, restarting"),
+                                                        Err(e) => {
+                                                            error!(error = %e, "me_drain_timeout_enforcer panicked, restarting in 1s");
+                                                            tokio::time::sleep(Duration::from_secs(1)).await;
+                                                        }
+                                                    }
+                                                }
+                                            });
+                                            let pool_watchdog = pool_bg.clone();
+                                            tokio::spawn(async move {
+                                                loop {
+                                                    let p = pool_watchdog.clone();
+                                                    let res = tokio::spawn(async move {
+                                                        crate::transport::middle_proxy::me_zombie_writer_watchdog(p).await;
+                                                    })
+                                                    .await;
+                                                    match res {
+                                                        Ok(()) => warn!("me_zombie_writer_watchdog exited unexpectedly, restarting"),
+                                                        Err(e) => {
+                                                            error!(error = %e, "me_zombie_writer_watchdog panicked, restarting in 1s");
+                                                            tokio::time::sleep(Duration::from_secs(1)).await;
+                                                        }
+                                                    }
+                                                }
+                                            });
+                                            // CRITICAL: keep the current-thread runtime
+                                            // alive. Without this, block_on() returns,
+                                            // the Runtime is dropped, and ALL spawned
+                                            // background tasks (health monitor, drain
+                                            // enforcer, zombie watchdog) are silently
+                                            // cancelled — causing the draining-writer
+                                            // leak that brought us here.
+                                            std::future::pending::<()>().await;
+                                            unreachable!();
                                     }
                                     Err(e) => {
                                         startup_tracker_bg.set_me_last_error(Some(e.to_string())).await;
@@ -408,22 +459,64 @@ pub(crate) async fn initialize_me_pool(
                                     "Middle-End pool initialized successfully"
                                 );
 
-                                let pool_clone = pool.clone();
+                                    // ── Supervised background tasks ──────────────────
-                                let rng_clone = rng.clone();
+                                    let pool_clone = pool.clone();
-                                let min_conns = pool_size;
+                                    let rng_clone = rng.clone();
-                                tokio::spawn(async move {
+                                    let min_conns = pool_size;
-                                    crate::transport::middle_proxy::me_health_monitor(
+                                    tokio::spawn(async move {
-                                        pool_clone, rng_clone, min_conns,
+                                        loop {
-                                    )
+                                            let p = pool_clone.clone();
-                                    .await;
+                                            let r = rng_clone.clone();
-                                });
+                                            let res = tokio::spawn(async move {
-                                let pool_drain_enforcer = pool.clone();
+                                                crate::transport::middle_proxy::me_health_monitor(
-                                tokio::spawn(async move {
+                                                    p, r, min_conns,
-                                    crate::transport::middle_proxy::me_drain_timeout_enforcer(
+                                                )
-                                        pool_drain_enforcer,
+                                                .await;
-                                    )
+                                            })
-                                    .await;
+                                            .await;
-                                });
+                                            match res {
+                                                Ok(()) => warn!("me_health_monitor exited unexpectedly, restarting"),
+                                                Err(e) => {
+                                                    error!(error = %e, "me_health_monitor panicked, restarting in 1s");
+                                                    tokio::time::sleep(Duration::from_secs(1)).await;
+                                                }
+                                            }
+                                        }
+                                    });
+                                    let pool_drain_enforcer = pool.clone();
+                                    tokio::spawn(async move {
+                                        loop {
+                                            let p = pool_drain_enforcer.clone();
+                                            let res = tokio::spawn(async move {
+                                                crate::transport::middle_proxy::me_drain_timeout_enforcer(p).await;
+                                            })
+                                            .await;
+                                            match res {
+                                                Ok(()) => warn!("me_drain_timeout_enforcer exited unexpectedly, restarting"),
+                                                Err(e) => {
+                                                    error!(error = %e, "me_drain_timeout_enforcer panicked, restarting in 1s");
+                                                    tokio::time::sleep(Duration::from_secs(1)).await;
+                                                }
+                                            }
+                                        }
+                                    });
+                                    let pool_watchdog = pool.clone();
+                                    tokio::spawn(async move {
+                                        loop {
+                                            let p = pool_watchdog.clone();
+                                            let res = tokio::spawn(async move {
+                                                crate::transport::middle_proxy::me_zombie_writer_watchdog(p).await;
+                                            })
+                                            .await;
+                                            match res {
+                                                Ok(()) => warn!("me_zombie_writer_watchdog exited unexpectedly, restarting"),
+                                                Err(e) => {
+                                                    error!(error = %e, "me_zombie_writer_watchdog panicked, restarting in 1s");
+                                                    tokio::time::sleep(Duration::from_secs(1)).await;
+                                                }
+                                            }
+                                        }
+                                    });
     
                                 break Some(pool);
                             }

src/maestro/tls_bootstrap.rs

@@ -38,12 +38,15 @@ pub(crate) async fn bootstrap_tls_front(
             .clone()
             .unwrap_or_else(|| config.censorship.tls_domain.clone());
         let mask_unix_sock = config.censorship.mask_unix_sock.clone();
+        let tls_fetch_scope = (!config.censorship.tls_fetch_scope.is_empty())
+            .then(|| config.censorship.tls_fetch_scope.clone());
         let fetch_timeout = Duration::from_secs(5);
 
         let cache_initial = cache.clone();
         let domains_initial = tls_domains.to_vec();
         let host_initial = mask_host.clone();
         let unix_sock_initial = mask_unix_sock.clone();
+        let scope_initial = tls_fetch_scope.clone();
         let upstream_initial = upstream_manager.clone();
         tokio::spawn(async move {
             let mut join = tokio::task::JoinSet::new();
@@ -51,6 +54,7 @@ pub(crate) async fn bootstrap_tls_front(
                 let cache_domain = cache_initial.clone();
                 let host_domain = host_initial.clone();
                 let unix_sock_domain = unix_sock_initial.clone();
+                let scope_domain = scope_initial.clone();
                 let upstream_domain = upstream_initial.clone();
                 join.spawn(async move {
                     match crate::tls_front::fetcher::fetch_real_tls(
@@ -59,6 +63,7 @@ pub(crate) async fn bootstrap_tls_front(
                         &domain,
                         fetch_timeout,
                         Some(upstream_domain),
+                        scope_domain.as_deref(),
                         proxy_protocol,
                         unix_sock_domain.as_deref(),
                     )
@@ -100,6 +105,7 @@ pub(crate) async fn bootstrap_tls_front(
         let domains_refresh = tls_domains.to_vec();
         let host_refresh = mask_host.clone();
         let unix_sock_refresh = mask_unix_sock.clone();
+        let scope_refresh = tls_fetch_scope.clone();
         let upstream_refresh = upstream_manager.clone();
         tokio::spawn(async move {
             loop {
@@ -112,6 +118,7 @@ pub(crate) async fn bootstrap_tls_front(
                     let cache_domain = cache_refresh.clone();
                     let host_domain = host_refresh.clone();
                     let unix_sock_domain = unix_sock_refresh.clone();
+                    let scope_domain = scope_refresh.clone();
                     let upstream_domain = upstream_refresh.clone();
                     join.spawn(async move {
                         match crate::tls_front::fetcher::fetch_real_tls(
@@ -120,6 +127,7 @@ pub(crate) async fn bootstrap_tls_front(
                             &domain,
                             fetch_timeout,
                             Some(upstream_domain),
+                            scope_domain.as_deref(),
                             proxy_protocol,
                             unix_sock_domain.as_deref(),
                         )

src/metrics.rs

@@ -16,7 +16,9 @@ use tracing::{info, warn, debug};
 use crate::config::ProxyConfig;
 use crate::ip_tracker::UserIpTracker;
 use crate::stats::beobachten::BeobachtenStore;
-use crate::stats::Stats;
+use crate::stats::{
+    MeWriterCleanupSideEffectStep, MeWriterTeardownMode, MeWriterTeardownReason, Stats,
+};
 use crate::transport::{ListenOptions, create_listener};
 
 pub async fn serve(
@@ -1770,6 +1772,169 @@ async fn render_metrics(stats: &Stats, config: &ProxyConfig, ip_tracker: &UserIp
         }
     );
 
+    let _ = writeln!(
+        out,
+        "# HELP telemt_me_writer_teardown_attempt_total ME writer teardown attempts by reason and mode"
+    );
+    let _ = writeln!(out, "# TYPE telemt_me_writer_teardown_attempt_total counter");
+    for reason in MeWriterTeardownReason::ALL {
+        for mode in MeWriterTeardownMode::ALL {
+            let _ = writeln!(
+                out,
+                "telemt_me_writer_teardown_attempt_total{{reason=\"{}\",mode=\"{}\"}} {}",
+                reason.as_str(),
+                mode.as_str(),
+                if me_allows_normal {
+                    stats.get_me_writer_teardown_attempt_total(reason, mode)
+                } else {
+                    0
+                }
+            );
+        }
+    }
+
+    let _ = writeln!(
+        out,
+        "# HELP telemt_me_writer_teardown_success_total ME writer teardown successes by mode"
+    );
+    let _ = writeln!(out, "# TYPE telemt_me_writer_teardown_success_total counter");
+    for mode in MeWriterTeardownMode::ALL {
+        let _ = writeln!(
+            out,
+            "telemt_me_writer_teardown_success_total{{mode=\"{}\"}} {}",
+            mode.as_str(),
+            if me_allows_normal {
+                stats.get_me_writer_teardown_success_total(mode)
+            } else {
+                0
+            }
+        );
+    }
+
+    let _ = writeln!(
+        out,
+        "# HELP telemt_me_writer_teardown_timeout_total Teardown operations that timed out"
+    );
+    let _ = writeln!(out, "# TYPE telemt_me_writer_teardown_timeout_total counter");
+    let _ = writeln!(
+        out,
+        "telemt_me_writer_teardown_timeout_total {}",
+        if me_allows_normal {
+            stats.get_me_writer_teardown_timeout_total()
+        } else {
+            0
+        }
+    );
+
+    let _ = writeln!(
+        out,
+        "# HELP telemt_me_writer_teardown_escalation_total Watchdog teardown escalations to hard detach"
+    );
+    let _ = writeln!(
+        out,
+        "# TYPE telemt_me_writer_teardown_escalation_total counter"
+    );
+    let _ = writeln!(
+        out,
+        "telemt_me_writer_teardown_escalation_total {}",
+        if me_allows_normal {
+            stats.get_me_writer_teardown_escalation_total()
+        } else {
+            0
+        }
+    );
+
+    let _ = writeln!(
+        out,
+        "# HELP telemt_me_writer_teardown_noop_total Teardown operations that became no-op"
+    );
+    let _ = writeln!(out, "# TYPE telemt_me_writer_teardown_noop_total counter");
+    let _ = writeln!(
+        out,
+        "telemt_me_writer_teardown_noop_total {}",
+        if me_allows_normal {
+            stats.get_me_writer_teardown_noop_total()
+        } else {
+            0
+        }
+    );
+
+    let _ = writeln!(
+        out,
+        "# HELP telemt_me_writer_teardown_duration_seconds ME writer teardown latency histogram by mode"
+    );
+    let _ = writeln!(
+        out,
+        "# TYPE telemt_me_writer_teardown_duration_seconds histogram"
+    );
+    let bucket_labels = Stats::me_writer_teardown_duration_bucket_labels();
+    for mode in MeWriterTeardownMode::ALL {
+        for (bucket_idx, label) in bucket_labels.iter().enumerate() {
+            let _ = writeln!(
+                out,
+                "telemt_me_writer_teardown_duration_seconds_bucket{{mode=\"{}\",le=\"{}\"}} {}",
+                mode.as_str(),
+                label,
+                if me_allows_normal {
+                    stats.get_me_writer_teardown_duration_bucket_total(mode, bucket_idx)
+                } else {
+                    0
+                }
+            );
+        }
+        let _ = writeln!(
+            out,
+            "telemt_me_writer_teardown_duration_seconds_bucket{{mode=\"{}\",le=\"+Inf\"}} {}",
+            mode.as_str(),
+            if me_allows_normal {
+                stats.get_me_writer_teardown_duration_count(mode)
+            } else {
+                0
+            }
+        );
+        let _ = writeln!(
+            out,
+            "telemt_me_writer_teardown_duration_seconds_sum{{mode=\"{}\"}} {:.6}",
+            mode.as_str(),
+            if me_allows_normal {
+                stats.get_me_writer_teardown_duration_sum_seconds(mode)
+            } else {
+                0.0
+            }
+        );
+        let _ = writeln!(
+            out,
+            "telemt_me_writer_teardown_duration_seconds_count{{mode=\"{}\"}} {}",
+            mode.as_str(),
+            if me_allows_normal {
+                stats.get_me_writer_teardown_duration_count(mode)
+            } else {
+                0
+            }
+        );
+    }
+
+    let _ = writeln!(
+        out,
+        "# HELP telemt_me_writer_cleanup_side_effect_failures_total Failed cleanup side effects by step"
+    );
+    let _ = writeln!(
+        out,
+        "# TYPE telemt_me_writer_cleanup_side_effect_failures_total counter"
+    );
+    for step in MeWriterCleanupSideEffectStep::ALL {
+        let _ = writeln!(
+            out,
+            "telemt_me_writer_cleanup_side_effect_failures_total{{step=\"{}\"}} {}",
+            step.as_str(),
+            if me_allows_normal {
+                stats.get_me_writer_cleanup_side_effect_failures_total(step)
+            } 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!(
@@ -2175,6 +2340,17 @@ mod tests {
         assert!(output.contains("# TYPE telemt_me_rpc_proxy_req_signal_sent_total counter"));
         assert!(output.contains("# TYPE telemt_me_idle_close_by_peer_total counter"));
         assert!(output.contains("# TYPE telemt_me_writer_removed_total counter"));
+        assert!(output.contains("# TYPE telemt_me_writer_teardown_attempt_total counter"));
+        assert!(output.contains("# TYPE telemt_me_writer_teardown_success_total counter"));
+        assert!(output.contains("# TYPE telemt_me_writer_teardown_timeout_total counter"));
+        assert!(output.contains("# TYPE telemt_me_writer_teardown_escalation_total counter"));
+        assert!(output.contains("# TYPE telemt_me_writer_teardown_noop_total counter"));
+        assert!(output.contains(
+            "# TYPE telemt_me_writer_teardown_duration_seconds histogram"
+        ));
+        assert!(output.contains(
+            "# TYPE telemt_me_writer_cleanup_side_effect_failures_total counter"
+        ));
         assert!(output.contains("# TYPE telemt_me_writer_close_signal_drop_total counter"));
         assert!(output.contains(
             "# TYPE telemt_me_writer_close_signal_channel_full_total counter"

src/proxy/direct_relay.rs

@@ -3,8 +3,7 @@ use std::io::Write;
 use std::net::SocketAddr;
 use std::sync::Arc;
 
-use tokio::io::{AsyncRead, AsyncWrite, AsyncWriteExt};
+use tokio::io::{AsyncRead, AsyncWrite, AsyncWriteExt, ReadHalf, WriteHalf, split};
-use tokio::net::TcpStream;
 use tokio::sync::watch;
 use tracing::{debug, info, warn};
 
@@ -15,7 +14,7 @@ use crate::protocol::constants::*;
 use crate::proxy::handshake::{HandshakeSuccess, encrypt_tg_nonce_with_ciphers, generate_tg_nonce};
 use crate::proxy::relay::relay_bidirectional;
 use crate::proxy::route_mode::{
-    RelayRouteMode, RouteCutoverState, ROUTE_SWITCH_ERROR_MSG, affected_cutover_state,
+    ROUTE_SWITCH_ERROR_MSG, RelayRouteMode, RouteCutoverState, affected_cutover_state,
     cutover_stagger_delay,
 };
 use crate::proxy::adaptive_buffers;
@@ -56,7 +55,11 @@ where
     );
 
     let tg_stream = upstream_manager
-        .connect(dc_addr, Some(success.dc_idx), user.strip_prefix("scope_").filter(|s| !s.is_empty()))
+        .connect(
+            dc_addr,
+            Some(success.dc_idx),
+            user.strip_prefix("scope_").filter(|s| !s.is_empty()),
+        )
         .await?;
 
     debug!(peer = %success.peer, dc_addr = %dc_addr, "Connected, performing TG handshake");
@@ -93,11 +96,9 @@ where
     );
     tokio::pin!(relay_result);
     let relay_result = loop {
-        if let Some(cutover) = affected_cutover_state(
+        if let Some(cutover) =
-            &route_rx,
+            affected_cutover_state(&route_rx, RelayRouteMode::Direct, route_snapshot.generation)
-            RelayRouteMode::Direct,
+        {
-            route_snapshot.generation,
-        ) {
             let delay = cutover_stagger_delay(session_id, cutover.generation);
             warn!(
                 user = %user,
@@ -148,7 +149,9 @@ fn get_dc_addr_static(dc_idx: i16, config: &ProxyConfig) -> Result<SocketAddr> {
         for addr_str in addrs {
             match addr_str.parse::<SocketAddr>() {
                 Ok(addr) => parsed.push(addr),
-                Err(_) => warn!(dc_idx = dc_idx, addr_str = %addr_str, "Invalid DC override address in config, ignoring"),
+                Err(_) => {
+                    warn!(dc_idx = dc_idx, addr_str = %addr_str, "Invalid DC override address in config, ignoring")
+                }
             }
         }
 
@@ -170,7 +173,10 @@ fn get_dc_addr_static(dc_idx: i16, config: &ProxyConfig) -> Result<SocketAddr> {
 
     // Unknown DC requested by client without override: log and fall back.
     if !config.dc_overrides.contains_key(&dc_key) {
-        warn!(dc_idx = dc_idx, "Requested non-standard DC with no override; falling back to default cluster");
+        warn!(
+            dc_idx = dc_idx,
+            "Requested non-standard DC with no override; falling back to default cluster"
+        );
         if config.general.unknown_dc_file_log_enabled
             && let Some(path) = &config.general.unknown_dc_log_path
             && let Ok(handle) = tokio::runtime::Handle::try_current()
@@ -204,15 +210,15 @@ fn get_dc_addr_static(dc_idx: i16, config: &ProxyConfig) -> Result<SocketAddr> {
     ))
 }
 
-async fn do_tg_handshake_static(
+async fn do_tg_handshake_static<S>(
-    mut stream: TcpStream,
+    mut stream: S,
     success: &HandshakeSuccess,
     config: &ProxyConfig,
     rng: &SecureRandom,
-) -> Result<(
+) -> Result<(CryptoReader<ReadHalf<S>>, CryptoWriter<WriteHalf<S>>)>
-    CryptoReader<tokio::net::tcp::OwnedReadHalf>,
+where
-    CryptoWriter<tokio::net::tcp::OwnedWriteHalf>,
+    S: AsyncRead + AsyncWrite + Unpin,
-)> {
+{
     let (nonce, _tg_enc_key, _tg_enc_iv, _tg_dec_key, _tg_dec_iv) = generate_tg_nonce(
         success.proto_tag,
         success.dc_idx,
@@ -235,7 +241,7 @@ async fn do_tg_handshake_static(
     stream.write_all(&encrypted_nonce).await?;
     stream.flush().await?;
 
-    let (read_half, write_half) = stream.into_split();
+    let (read_half, write_half) = split(stream);
 
     let max_pending = config.general.crypto_pending_buffer;
     Ok((

src/stats/mod.rs

@@ -19,6 +19,137 @@ use tracing::debug;
 use crate::config::{MeTelemetryLevel, MeWriterPickMode};
 use self::telemetry::TelemetryPolicy;
 
+const ME_WRITER_TEARDOWN_MODE_COUNT: usize = 2;
+const ME_WRITER_TEARDOWN_REASON_COUNT: usize = 11;
+const ME_WRITER_CLEANUP_SIDE_EFFECT_STEP_COUNT: usize = 2;
+const ME_WRITER_TEARDOWN_DURATION_BUCKET_COUNT: usize = 12;
+const ME_WRITER_TEARDOWN_DURATION_BUCKET_BOUNDS_MICROS: [u64; ME_WRITER_TEARDOWN_DURATION_BUCKET_COUNT] = [
+    1_000,
+    5_000,
+    10_000,
+    25_000,
+    50_000,
+    100_000,
+    250_000,
+    500_000,
+    1_000_000,
+    2_500_000,
+    5_000_000,
+    10_000_000,
+];
+const ME_WRITER_TEARDOWN_DURATION_BUCKET_LABELS: [&str; ME_WRITER_TEARDOWN_DURATION_BUCKET_COUNT] = [
+    "0.001",
+    "0.005",
+    "0.01",
+    "0.025",
+    "0.05",
+    "0.1",
+    "0.25",
+    "0.5",
+    "1",
+    "2.5",
+    "5",
+    "10",
+];
+
+#[derive(Clone, Copy, Debug, PartialEq, Eq, Hash)]
+#[repr(u8)]
+pub enum MeWriterTeardownMode {
+    Normal = 0,
+    HardDetach = 1,
+}
+
+impl MeWriterTeardownMode {
+    pub const ALL: [Self; ME_WRITER_TEARDOWN_MODE_COUNT] =
+        [Self::Normal, Self::HardDetach];
+
+    pub const fn as_str(self) -> &'static str {
+        match self {
+            Self::Normal => "normal",
+            Self::HardDetach => "hard_detach",
+        }
+    }
+
+    const fn idx(self) -> usize {
+        self as usize
+    }
+}
+
+#[derive(Clone, Copy, Debug, PartialEq, Eq, Hash)]
+#[repr(u8)]
+pub enum MeWriterTeardownReason {
+    ReaderExit = 0,
+    WriterTaskExit = 1,
+    PingSendFail = 2,
+    SignalSendFail = 3,
+    RouteChannelClosed = 4,
+    CloseRpcChannelClosed = 5,
+    PruneClosedWriter = 6,
+    ReapTimeoutExpired = 7,
+    ReapThresholdForce = 8,
+    ReapEmpty = 9,
+    WatchdogStuckDraining = 10,
+}
+
+impl MeWriterTeardownReason {
+    pub const ALL: [Self; ME_WRITER_TEARDOWN_REASON_COUNT] = [
+        Self::ReaderExit,
+        Self::WriterTaskExit,
+        Self::PingSendFail,
+        Self::SignalSendFail,
+        Self::RouteChannelClosed,
+        Self::CloseRpcChannelClosed,
+        Self::PruneClosedWriter,
+        Self::ReapTimeoutExpired,
+        Self::ReapThresholdForce,
+        Self::ReapEmpty,
+        Self::WatchdogStuckDraining,
+    ];
+
+    pub const fn as_str(self) -> &'static str {
+        match self {
+            Self::ReaderExit => "reader_exit",
+            Self::WriterTaskExit => "writer_task_exit",
+            Self::PingSendFail => "ping_send_fail",
+            Self::SignalSendFail => "signal_send_fail",
+            Self::RouteChannelClosed => "route_channel_closed",
+            Self::CloseRpcChannelClosed => "close_rpc_channel_closed",
+            Self::PruneClosedWriter => "prune_closed_writer",
+            Self::ReapTimeoutExpired => "reap_timeout_expired",
+            Self::ReapThresholdForce => "reap_threshold_force",
+            Self::ReapEmpty => "reap_empty",
+            Self::WatchdogStuckDraining => "watchdog_stuck_draining",
+        }
+    }
+
+    const fn idx(self) -> usize {
+        self as usize
+    }
+}
+
+#[derive(Clone, Copy, Debug, PartialEq, Eq, Hash)]
+#[repr(u8)]
+pub enum MeWriterCleanupSideEffectStep {
+    CloseSignalChannelFull = 0,
+    CloseSignalChannelClosed = 1,
+}
+
+impl MeWriterCleanupSideEffectStep {
+    pub const ALL: [Self; ME_WRITER_CLEANUP_SIDE_EFFECT_STEP_COUNT] =
+        [Self::CloseSignalChannelFull, Self::CloseSignalChannelClosed];
+
+    pub const fn as_str(self) -> &'static str {
+        match self {
+            Self::CloseSignalChannelFull => "close_signal_channel_full",
+            Self::CloseSignalChannelClosed => "close_signal_channel_closed",
+        }
+    }
+
+    const fn idx(self) -> usize {
+        self as usize
+    }
+}
+
 // ============= Stats =============
 
 #[derive(Default)]
@@ -128,6 +259,18 @@ pub struct Stats {
     me_draining_writers_reap_progress_total: AtomicU64,
     me_writer_removed_total: AtomicU64,
     me_writer_removed_unexpected_total: AtomicU64,
+    me_writer_teardown_attempt_total:
+        [[AtomicU64; ME_WRITER_TEARDOWN_MODE_COUNT]; ME_WRITER_TEARDOWN_REASON_COUNT],
+    me_writer_teardown_success_total: [AtomicU64; ME_WRITER_TEARDOWN_MODE_COUNT],
+    me_writer_teardown_timeout_total: AtomicU64,
+    me_writer_teardown_escalation_total: AtomicU64,
+    me_writer_teardown_noop_total: AtomicU64,
+    me_writer_cleanup_side_effect_failures_total:
+        [AtomicU64; ME_WRITER_CLEANUP_SIDE_EFFECT_STEP_COUNT],
+    me_writer_teardown_duration_bucket_hits:
+        [[AtomicU64; ME_WRITER_TEARDOWN_DURATION_BUCKET_COUNT + 1]; ME_WRITER_TEARDOWN_MODE_COUNT],
+    me_writer_teardown_duration_sum_micros: [AtomicU64; ME_WRITER_TEARDOWN_MODE_COUNT],
+    me_writer_teardown_duration_count: [AtomicU64; ME_WRITER_TEARDOWN_MODE_COUNT],
     me_refill_triggered_total: AtomicU64,
     me_refill_skipped_inflight_total: AtomicU64,
     me_refill_failed_total: AtomicU64,
@@ -765,6 +908,74 @@ impl Stats {
             self.me_writer_removed_unexpected_total.fetch_add(1, Ordering::Relaxed);
         }
     }
+    pub fn increment_me_writer_teardown_attempt_total(
+        &self,
+        reason: MeWriterTeardownReason,
+        mode: MeWriterTeardownMode,
+    ) {
+        if self.telemetry_me_allows_normal() {
+            self.me_writer_teardown_attempt_total[reason.idx()][mode.idx()]
+                .fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_writer_teardown_success_total(&self, mode: MeWriterTeardownMode) {
+        if self.telemetry_me_allows_normal() {
+            self.me_writer_teardown_success_total[mode.idx()].fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_writer_teardown_timeout_total(&self) {
+        if self.telemetry_me_allows_normal() {
+            self.me_writer_teardown_timeout_total
+                .fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_writer_teardown_escalation_total(&self) {
+        if self.telemetry_me_allows_normal() {
+            self.me_writer_teardown_escalation_total
+                .fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_writer_teardown_noop_total(&self) {
+        if self.telemetry_me_allows_normal() {
+            self.me_writer_teardown_noop_total
+                .fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn increment_me_writer_cleanup_side_effect_failures_total(
+        &self,
+        step: MeWriterCleanupSideEffectStep,
+    ) {
+        if self.telemetry_me_allows_normal() {
+            self.me_writer_cleanup_side_effect_failures_total[step.idx()]
+                .fetch_add(1, Ordering::Relaxed);
+        }
+    }
+    pub fn observe_me_writer_teardown_duration(
+        &self,
+        mode: MeWriterTeardownMode,
+        duration: Duration,
+    ) {
+        if !self.telemetry_me_allows_normal() {
+            return;
+        }
+        let duration_micros = duration.as_micros().min(u64::MAX as u128) as u64;
+        let mut bucket_idx = ME_WRITER_TEARDOWN_DURATION_BUCKET_COUNT;
+        for (idx, upper_bound_micros) in ME_WRITER_TEARDOWN_DURATION_BUCKET_BOUNDS_MICROS
+            .iter()
+            .copied()
+            .enumerate()
+        {
+            if duration_micros <= upper_bound_micros {
+                bucket_idx = idx;
+                break;
+            }
+        }
+        self.me_writer_teardown_duration_bucket_hits[mode.idx()][bucket_idx]
+            .fetch_add(1, Ordering::Relaxed);
+        self.me_writer_teardown_duration_sum_micros[mode.idx()]
+            .fetch_add(duration_micros, Ordering::Relaxed);
+        self.me_writer_teardown_duration_count[mode.idx()].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);
@@ -1297,6 +1508,79 @@ impl Stats {
     pub fn get_me_writer_removed_unexpected_total(&self) -> u64 {
         self.me_writer_removed_unexpected_total.load(Ordering::Relaxed)
     }
+    pub fn get_me_writer_teardown_attempt_total(
+        &self,
+        reason: MeWriterTeardownReason,
+        mode: MeWriterTeardownMode,
+    ) -> u64 {
+        self.me_writer_teardown_attempt_total[reason.idx()][mode.idx()]
+            .load(Ordering::Relaxed)
+    }
+    pub fn get_me_writer_teardown_attempt_total_by_mode(&self, mode: MeWriterTeardownMode) -> u64 {
+        MeWriterTeardownReason::ALL
+            .iter()
+            .copied()
+            .map(|reason| self.get_me_writer_teardown_attempt_total(reason, mode))
+            .sum()
+    }
+    pub fn get_me_writer_teardown_success_total(&self, mode: MeWriterTeardownMode) -> u64 {
+        self.me_writer_teardown_success_total[mode.idx()].load(Ordering::Relaxed)
+    }
+    pub fn get_me_writer_teardown_timeout_total(&self) -> u64 {
+        self.me_writer_teardown_timeout_total.load(Ordering::Relaxed)
+    }
+    pub fn get_me_writer_teardown_escalation_total(&self) -> u64 {
+        self.me_writer_teardown_escalation_total
+            .load(Ordering::Relaxed)
+    }
+    pub fn get_me_writer_teardown_noop_total(&self) -> u64 {
+        self.me_writer_teardown_noop_total.load(Ordering::Relaxed)
+    }
+    pub fn get_me_writer_cleanup_side_effect_failures_total(
+        &self,
+        step: MeWriterCleanupSideEffectStep,
+    ) -> u64 {
+        self.me_writer_cleanup_side_effect_failures_total[step.idx()]
+            .load(Ordering::Relaxed)
+    }
+    pub fn get_me_writer_cleanup_side_effect_failures_total_all(&self) -> u64 {
+        MeWriterCleanupSideEffectStep::ALL
+            .iter()
+            .copied()
+            .map(|step| self.get_me_writer_cleanup_side_effect_failures_total(step))
+            .sum()
+    }
+    pub fn me_writer_teardown_duration_bucket_labels(
+    ) -> &'static [&'static str; ME_WRITER_TEARDOWN_DURATION_BUCKET_COUNT] {
+        &ME_WRITER_TEARDOWN_DURATION_BUCKET_LABELS
+    }
+    pub fn get_me_writer_teardown_duration_bucket_hits(
+        &self,
+        mode: MeWriterTeardownMode,
+        bucket_idx: usize,
+    ) -> u64 {
+        self.me_writer_teardown_duration_bucket_hits[mode.idx()][bucket_idx]
+            .load(Ordering::Relaxed)
+    }
+    pub fn get_me_writer_teardown_duration_bucket_total(
+        &self,
+        mode: MeWriterTeardownMode,
+        bucket_idx: usize,
+    ) -> u64 {
+        let capped_idx = bucket_idx.min(ME_WRITER_TEARDOWN_DURATION_BUCKET_COUNT);
+        let mut total = 0u64;
+        for idx in 0..=capped_idx {
+            total = total.saturating_add(self.get_me_writer_teardown_duration_bucket_hits(mode, idx));
+        }
+        total
+    }
+    pub fn get_me_writer_teardown_duration_count(&self, mode: MeWriterTeardownMode) -> u64 {
+        self.me_writer_teardown_duration_count[mode.idx()].load(Ordering::Relaxed)
+    }
+    pub fn get_me_writer_teardown_duration_sum_seconds(&self, mode: MeWriterTeardownMode) -> f64 {
+        self.me_writer_teardown_duration_sum_micros[mode.idx()].load(Ordering::Relaxed) as f64
+            / 1_000_000.0
+    }
     pub fn get_me_refill_triggered_total(&self) -> u64 {
         self.me_refill_triggered_total.load(Ordering::Relaxed)
     }
@@ -1801,6 +2085,79 @@ mod tests {
         assert_eq!(stats.get_me_route_drop_queue_full(), 0);
     }
 
+    #[test]
+    fn test_teardown_counters_and_duration() {
+        let stats = Stats::new();
+        stats.increment_me_writer_teardown_attempt_total(
+            MeWriterTeardownReason::ReaderExit,
+            MeWriterTeardownMode::Normal,
+        );
+        stats.increment_me_writer_teardown_success_total(MeWriterTeardownMode::Normal);
+        stats.observe_me_writer_teardown_duration(
+            MeWriterTeardownMode::Normal,
+            Duration::from_millis(3),
+        );
+        stats.increment_me_writer_cleanup_side_effect_failures_total(
+            MeWriterCleanupSideEffectStep::CloseSignalChannelFull,
+        );
+
+        assert_eq!(
+            stats.get_me_writer_teardown_attempt_total(
+                MeWriterTeardownReason::ReaderExit,
+                MeWriterTeardownMode::Normal
+            ),
+            1
+        );
+        assert_eq!(
+            stats.get_me_writer_teardown_success_total(MeWriterTeardownMode::Normal),
+            1
+        );
+        assert_eq!(
+            stats.get_me_writer_teardown_duration_count(MeWriterTeardownMode::Normal),
+            1
+        );
+        assert!(
+            stats.get_me_writer_teardown_duration_sum_seconds(MeWriterTeardownMode::Normal) > 0.0
+        );
+        assert_eq!(
+            stats.get_me_writer_cleanup_side_effect_failures_total(
+                MeWriterCleanupSideEffectStep::CloseSignalChannelFull
+            ),
+            1
+        );
+    }
+
+    #[test]
+    fn test_teardown_counters_respect_me_silent() {
+        let stats = Stats::new();
+        stats.apply_telemetry_policy(TelemetryPolicy {
+            core_enabled: true,
+            user_enabled: true,
+            me_level: MeTelemetryLevel::Silent,
+        });
+        stats.increment_me_writer_teardown_attempt_total(
+            MeWriterTeardownReason::ReaderExit,
+            MeWriterTeardownMode::Normal,
+        );
+        stats.increment_me_writer_teardown_timeout_total();
+        stats.observe_me_writer_teardown_duration(
+            MeWriterTeardownMode::Normal,
+            Duration::from_millis(1),
+        );
+        assert_eq!(
+            stats.get_me_writer_teardown_attempt_total(
+                MeWriterTeardownReason::ReaderExit,
+                MeWriterTeardownMode::Normal
+            ),
+            0
+        );
+        assert_eq!(stats.get_me_writer_teardown_timeout_total(), 0);
+        assert_eq!(
+            stats.get_me_writer_teardown_duration_count(MeWriterTeardownMode::Normal),
+            0
+        );
+    }
+    
     #[test]
     fn test_replay_checker_basic() {
         let checker = ReplayChecker::new(100, Duration::from_secs(60));

src/tls_front/fetcher.rs

@@ -7,33 +7,29 @@ use tokio::net::TcpStream;
 #[cfg(unix)]
 use tokio::net::UnixStream;
 use tokio::time::timeout;
-use tokio_rustls::client::TlsStream;
 use tokio_rustls::TlsConnector;
+use tokio_rustls::client::TlsStream;
 use tracing::{debug, warn};
 
-use rustls::client::danger::{HandshakeSignatureValid, ServerCertVerified, ServerCertVerifier};
 use rustls::client::ClientConfig;
+use rustls::client::danger::{HandshakeSignatureValid, ServerCertVerified, ServerCertVerifier};
 use rustls::pki_types::{CertificateDer, ServerName, UnixTime};
 use rustls::{DigitallySignedStruct, Error as RustlsError};
 
-use x509_parser::prelude::FromDer;
 use x509_parser::certificate::X509Certificate;
+use x509_parser::prelude::FromDer;
 
 use crate::crypto::SecureRandom;
 use crate::network::dns_overrides::resolve_socket_addr;
 use crate::protocol::constants::{
     TLS_RECORD_APPLICATION, TLS_RECORD_CHANGE_CIPHER, TLS_RECORD_HANDSHAKE,
 };
-use crate::transport::proxy_protocol::{ProxyProtocolV1Builder, ProxyProtocolV2Builder};
 use crate::tls_front::types::{
-    ParsedCertificateInfo,
+    ParsedCertificateInfo, ParsedServerHello, TlsBehaviorProfile, TlsCertPayload, TlsExtension,
-    ParsedServerHello,
+    TlsFetchResult, TlsProfileSource,
-    TlsBehaviorProfile,
-    TlsCertPayload,
-    TlsExtension,
-    TlsFetchResult,
-    TlsProfileSource,
 };
+use crate::transport::UpstreamStream;
+use crate::transport::proxy_protocol::{ProxyProtocolV1Builder, ProxyProtocolV2Builder};
 
 /// No-op verifier: accept any certificate (we only need lengths and metadata).
 #[derive(Debug)]
@@ -144,21 +140,27 @@ fn build_client_hello(sni: &str, rng: &SecureRandom) -> Vec<u8> {
     exts.extend_from_slice(&0x000au16.to_be_bytes());
     exts.extend_from_slice(&((2 + groups.len() * 2) as u16).to_be_bytes());
     exts.extend_from_slice(&(groups.len() as u16 * 2).to_be_bytes());
-    for g in groups { exts.extend_from_slice(&g.to_be_bytes()); }
+    for g in groups {
+        exts.extend_from_slice(&g.to_be_bytes());
+    }
 
     // signature_algorithms
     let sig_algs: [u16; 4] = [0x0804, 0x0805, 0x0403, 0x0503]; // rsa_pss_rsae_sha256/384, ecdsa_secp256r1_sha256, rsa_pkcs1_sha256
     exts.extend_from_slice(&0x000du16.to_be_bytes());
     exts.extend_from_slice(&((2 + sig_algs.len() * 2) as u16).to_be_bytes());
     exts.extend_from_slice(&(sig_algs.len() as u16 * 2).to_be_bytes());
-    for a in sig_algs { exts.extend_from_slice(&a.to_be_bytes()); }
+    for a in sig_algs {
+        exts.extend_from_slice(&a.to_be_bytes());
+    }
 
     // supported_versions (TLS1.3 + TLS1.2)
     let versions: [u16; 2] = [0x0304, 0x0303];
     exts.extend_from_slice(&0x002bu16.to_be_bytes());
     exts.extend_from_slice(&((1 + versions.len() * 2) as u16).to_be_bytes());
     exts.push((versions.len() * 2) as u8);
-    for v in versions { exts.extend_from_slice(&v.to_be_bytes()); }
+    for v in versions {
+        exts.extend_from_slice(&v.to_be_bytes());
+    }
 
     // key_share (x25519)
     let key = gen_key_share(rng);
@@ -273,7 +275,10 @@ fn parse_server_hello(body: &[u8]) -> Option<ParsedServerHello> {
         pos += 4;
         let data = body.get(pos..pos + elen)?.to_vec();
         pos += elen;
-        extensions.push(TlsExtension { ext_type: etype, data });
+        extensions.push(TlsExtension {
+            ext_type: etype,
+            data,
+        });
     }
 
     Some(ParsedServerHello {
@@ -394,37 +399,42 @@ async fn connect_tcp_with_upstream(
     port: u16,
     connect_timeout: Duration,
     upstream: Option<std::sync::Arc<crate::transport::UpstreamManager>>,
-) -> Result<TcpStream> {
+    scope: Option<&str>,
+) -> Result<UpstreamStream> {
     if let Some(manager) = upstream {
         if let Some(addr) = resolve_socket_addr(host, port) {
-            match manager.connect(addr, None, None).await {
+            match manager.connect(addr, None, scope).await {
                 Ok(stream) => return Ok(stream),
                 Err(e) => {
                     warn!(
                         host = %host,
                         port = port,
+                        scope = ?scope,
                         error = %e,
                         "Upstream connect failed, using direct connect"
                     );
                 }
             }
-        } else if let Ok(mut addrs) = tokio::net::lookup_host((host, port)).await {
+        } else if let Ok(mut addrs) = tokio::net::lookup_host((host, port)).await
-            if let Some(addr) = addrs.find(|a| a.is_ipv4()) {
+            && let Some(addr) = addrs.find(|a| a.is_ipv4())
-                match manager.connect(addr, None, None).await {
+        {
-                    Ok(stream) => return Ok(stream),
+            match manager.connect(addr, None, scope).await {
-                    Err(e) => {
+                Ok(stream) => return Ok(stream),
-                        warn!(
+                Err(e) => {
-                            host = %host,
+                    warn!(
-                            port = port,
+                        host = %host,
-                            error = %e,
+                        port = port,
-                            "Upstream connect failed, using direct connect"
+                        scope = ?scope,
-                        );
+                        error = %e,
-                    }
+                        "Upstream connect failed, using direct connect"
+                    );
                 }
             }
         }
     }
-    connect_with_dns_override(host, port, connect_timeout).await
+    Ok(UpstreamStream::Tcp(
+        connect_with_dns_override(host, port, connect_timeout).await?,
+    ))
 }
 
 fn encode_tls13_certificate_message(cert_chain_der: &[Vec<u8>]) -> Option<Vec<u8>> {
@@ -443,9 +453,7 @@ fn encode_tls13_certificate_message(cert_chain_der: &[Vec<u8>]) -> Option<Vec<u8
     }
 
     // Certificate = context_len(1) + certificate_list_len(3) + entries
-    let body_len = 1usize
+    let body_len = 1usize.checked_add(3)?.checked_add(certificate_list.len())?;
-        .checked_add(3)?
-        .checked_add(certificate_list.len())?;
 
     let mut message = Vec::with_capacity(4 + body_len);
     message.push(0x0b); // HandshakeType::certificate
@@ -537,6 +545,7 @@ async fn fetch_via_raw_tls(
     sni: &str,
     connect_timeout: Duration,
     upstream: Option<std::sync::Arc<crate::transport::UpstreamManager>>,
+    scope: Option<&str>,
     proxy_protocol: u8,
     unix_sock: Option<&str>,
 ) -> Result<TlsFetchResult> {
@@ -549,7 +558,8 @@ async fn fetch_via_raw_tls(
                     sock = %sock_path,
                     "Raw TLS fetch using mask unix socket"
                 );
-                return fetch_via_raw_tls_stream(stream, sni, connect_timeout, proxy_protocol).await;
+                return fetch_via_raw_tls_stream(stream, sni, connect_timeout, proxy_protocol)
+                    .await;
             }
             Ok(Err(e)) => {
                 warn!(
@@ -572,7 +582,7 @@ async fn fetch_via_raw_tls(
     #[cfg(not(unix))]
     let _ = unix_sock;
 
-    let stream = connect_tcp_with_upstream(host, port, connect_timeout, upstream).await?;
+    let stream = connect_tcp_with_upstream(host, port, connect_timeout, upstream, scope).await?;
     fetch_via_raw_tls_stream(stream, sni, connect_timeout, proxy_protocol).await
 }
 
@@ -616,12 +626,13 @@ where
         .map(|slice| slice.to_vec())
         .unwrap_or_default();
     let cert_chain_der: Vec<Vec<u8>> = certs.iter().map(|c| c.as_ref().to_vec()).collect();
-    let cert_payload = encode_tls13_certificate_message(&cert_chain_der).map(|certificate_message| {
+    let cert_payload =
-        TlsCertPayload {
+        encode_tls13_certificate_message(&cert_chain_der).map(|certificate_message| {
-            cert_chain_der: cert_chain_der.clone(),
+            TlsCertPayload {
-            certificate_message,
+                cert_chain_der: cert_chain_der.clone(),
-        }
+                certificate_message,
-    });
+            }
+        });
 
     let total_cert_len = cert_payload
         .as_ref()
@@ -675,6 +686,7 @@ async fn fetch_via_rustls(
     sni: &str,
     connect_timeout: Duration,
     upstream: Option<std::sync::Arc<crate::transport::UpstreamManager>>,
+    scope: Option<&str>,
     proxy_protocol: u8,
     unix_sock: Option<&str>,
 ) -> Result<TlsFetchResult> {
@@ -710,7 +722,7 @@ async fn fetch_via_rustls(
     #[cfg(not(unix))]
     let _ = unix_sock;
 
-    let stream = connect_tcp_with_upstream(host, port, connect_timeout, upstream).await?;
+    let stream = connect_tcp_with_upstream(host, port, connect_timeout, upstream, scope).await?;
     fetch_via_rustls_stream(stream, host, sni, proxy_protocol).await
 }
 
@@ -726,6 +738,7 @@ pub async fn fetch_real_tls(
     sni: &str,
     connect_timeout: Duration,
     upstream: Option<std::sync::Arc<crate::transport::UpstreamManager>>,
+    scope: Option<&str>,
     proxy_protocol: u8,
     unix_sock: Option<&str>,
 ) -> Result<TlsFetchResult> {
@@ -735,6 +748,7 @@ pub async fn fetch_real_tls(
         sni,
         connect_timeout,
         upstream.clone(),
+        scope,
         proxy_protocol,
         unix_sock,
     )
@@ -753,6 +767,7 @@ pub async fn fetch_real_tls(
         sni,
         connect_timeout,
         upstream,
+        scope,
         proxy_protocol,
         unix_sock,
     )

src/transport/middle_proxy/health.rs

@@ -10,9 +10,10 @@ use tracing::{debug, info, warn};
 use crate::config::MeFloorMode;
 use crate::crypto::SecureRandom;
 use crate::network::IpFamily;
+use crate::stats::MeWriterTeardownReason;
 
 use super::MePool;
-use super::pool::MeWriter;
+use super::pool::{MeFamilyRuntimeState, MeWriter};
 
 const JITTER_FRAC_NUM: u64 = 2; // jitter up to 50% of backoff
 #[allow(dead_code)]
@@ -33,6 +34,33 @@ const HEALTH_DRAIN_SOFT_EVICT_BUDGET_MIN: usize = 8;
 const HEALTH_DRAIN_SOFT_EVICT_BUDGET_MAX: usize = 256;
 const HEALTH_DRAIN_REAP_OPPORTUNISTIC_INTERVAL_SECS: u64 = 1;
 const HEALTH_DRAIN_TIMEOUT_ENFORCER_INTERVAL_SECS: u64 = 1;
+const FAMILY_SUPPRESS_FAIL_STREAK_THRESHOLD: u32 = 6;
+const FAMILY_SUPPRESS_WINDOW_SECS: u64 = 120;
+const FAMILY_RECOVER_PROBE_INTERVAL_SECS: u64 = 5;
+const FAMILY_RECOVER_SUCCESS_STREAK_REQUIRED: u32 = 3;
+
+#[derive(Debug, Clone)]
+struct FamilyCircuitState {
+    state: MeFamilyRuntimeState,
+    state_since_at: Instant,
+    suppressed_until: Option<Instant>,
+    next_probe_at: Instant,
+    fail_streak: u32,
+    recover_success_streak: u32,
+}
+
+impl FamilyCircuitState {
+    fn new(now: Instant) -> Self {
+        Self {
+            state: MeFamilyRuntimeState::Healthy,
+            state_since_at: now,
+            suppressed_until: None,
+            next_probe_at: now,
+            fail_streak: 0,
+            recover_success_streak: 0,
+        }
+    }
+}
 
 #[derive(Debug, Clone)]
 struct DcFloorPlanEntry {
@@ -72,6 +100,25 @@ pub async fn me_health_monitor(pool: Arc<MePool>, rng: Arc<SecureRandom>, _min_c
     let mut floor_warn_next_allowed: HashMap<(i32, IpFamily), Instant> = HashMap::new();
     let mut drain_warn_next_allowed: HashMap<u64, Instant> = HashMap::new();
     let mut drain_soft_evict_next_allowed: HashMap<u64, Instant> = HashMap::new();
+    let mut family_v4_circuit = FamilyCircuitState::new(Instant::now());
+    let mut family_v6_circuit = FamilyCircuitState::new(Instant::now());
+    let init_epoch_secs = MePool::now_epoch_secs();
+    pool.set_family_runtime_state(
+        IpFamily::V4,
+        family_v4_circuit.state,
+        init_epoch_secs,
+        0,
+        family_v4_circuit.fail_streak,
+        family_v4_circuit.recover_success_streak,
+    );
+    pool.set_family_runtime_state(
+        IpFamily::V6,
+        family_v6_circuit.state,
+        init_epoch_secs,
+        0,
+        family_v6_circuit.fail_streak,
+        family_v6_circuit.recover_success_streak,
+    );
     let mut degraded_interval = true;
     loop {
         let interval = if degraded_interval {
@@ -87,7 +134,9 @@ pub async fn me_health_monitor(pool: Arc<MePool>, rng: Arc<SecureRandom>, _min_c
             &mut drain_soft_evict_next_allowed,
         )
         .await;
-        let v4_degraded = check_family(
+        let now = Instant::now();
+        let now_epoch_secs = MePool::now_epoch_secs();
+        let v4_degraded_raw = check_family(
             IpFamily::V4,
             &pool,
             &rng,
@@ -106,25 +155,53 @@ pub async fn me_health_monitor(pool: Arc<MePool>, rng: Arc<SecureRandom>, _min_c
             &mut drain_soft_evict_next_allowed,
         )
         .await;
-        let v6_degraded = check_family(
+        let v4_degraded = apply_family_circuit_result(
-            IpFamily::V6,
             &pool,
-            &rng,
+            IpFamily::V4,
-            &mut backoff,
+            &mut family_v4_circuit,
-            &mut next_attempt,
+            Some(v4_degraded_raw),
-            &mut inflight,
+            false,
-            &mut outage_backoff,
+            now,
-            &mut outage_next_attempt,
+            now_epoch_secs,
-            &mut single_endpoint_outage,
+        );
-            &mut shadow_rotate_deadline,
+
-            &mut idle_refresh_next_attempt,
+        let v6_check_ran = should_run_family_check(&mut family_v6_circuit, now);
-            &mut adaptive_idle_since,
+        let v6_degraded_raw = if v6_check_ran {
-            &mut adaptive_recover_until,
+            check_family(
-            &mut floor_warn_next_allowed,
+                IpFamily::V6,
-            &mut drain_warn_next_allowed,
+                &pool,
-            &mut drain_soft_evict_next_allowed,
+                &rng,
-        )
+                &mut backoff,
-        .await;
+                &mut next_attempt,
+                &mut inflight,
+                &mut outage_backoff,
+                &mut outage_next_attempt,
+                &mut single_endpoint_outage,
+                &mut shadow_rotate_deadline,
+                &mut idle_refresh_next_attempt,
+                &mut adaptive_idle_since,
+                &mut adaptive_recover_until,
+                &mut floor_warn_next_allowed,
+                &mut drain_warn_next_allowed,
+                &mut drain_soft_evict_next_allowed,
+            )
+            .await
+        } else {
+            false
+        };
+        let v6_degraded = apply_family_circuit_result(
+            &pool,
+            IpFamily::V6,
+            &mut family_v6_circuit,
+            if v6_check_ran {
+                Some(v6_degraded_raw)
+            } else {
+                None
+            },
+            true,
+            now,
+            now_epoch_secs,
+        );
         degraded_interval = v4_degraded || v6_degraded;
     }
 }
@@ -146,6 +223,148 @@ pub async fn me_drain_timeout_enforcer(pool: Arc<MePool>) {
     }
 }
 
+fn should_run_family_check(circuit: &mut FamilyCircuitState, now: Instant) -> bool {
+    match circuit.state {
+        MeFamilyRuntimeState::Suppressed => {
+            if now < circuit.next_probe_at {
+                return false;
+            }
+            circuit.next_probe_at =
+                now + Duration::from_secs(FAMILY_RECOVER_PROBE_INTERVAL_SECS);
+            true
+        }
+        _ => true,
+    }
+}
+
+fn apply_family_circuit_result(
+    pool: &Arc<MePool>,
+    family: IpFamily,
+    circuit: &mut FamilyCircuitState,
+    degraded: Option<bool>,
+    allow_suppress: bool,
+    now: Instant,
+    now_epoch_secs: u64,
+) -> bool {
+    let Some(degraded) = degraded else {
+        // Preserve suppression state when probe tick is intentionally skipped.
+        return false;
+    };
+
+    let previous_state = circuit.state;
+    match circuit.state {
+        MeFamilyRuntimeState::Suppressed => {
+            if degraded {
+                circuit.fail_streak = circuit.fail_streak.saturating_add(1);
+                circuit.recover_success_streak = 0;
+                let until = now + Duration::from_secs(FAMILY_SUPPRESS_WINDOW_SECS);
+                circuit.suppressed_until = Some(until);
+                circuit.state_since_at = now;
+                warn!(
+                    ?family,
+                    fail_streak = circuit.fail_streak,
+                    suppress_secs = FAMILY_SUPPRESS_WINDOW_SECS,
+                    "ME family remains suppressed due to ongoing failures"
+                );
+            } else {
+                circuit.fail_streak = 0;
+                circuit.recover_success_streak = 1;
+                circuit.state = MeFamilyRuntimeState::Recovering;
+            }
+        }
+        MeFamilyRuntimeState::Recovering => {
+            if degraded {
+                circuit.fail_streak = circuit.fail_streak.saturating_add(1);
+                if allow_suppress {
+                    circuit.state = MeFamilyRuntimeState::Suppressed;
+                    let until = now + Duration::from_secs(FAMILY_SUPPRESS_WINDOW_SECS);
+                    circuit.suppressed_until = Some(until);
+                    circuit.next_probe_at =
+                        now + Duration::from_secs(FAMILY_RECOVER_PROBE_INTERVAL_SECS);
+                    warn!(
+                        ?family,
+                        fail_streak = circuit.fail_streak,
+                        suppress_secs = FAMILY_SUPPRESS_WINDOW_SECS,
+                        "ME family temporarily suppressed after repeated degradation"
+                    );
+                } else {
+                    circuit.state = MeFamilyRuntimeState::Degraded;
+                }
+            } else {
+                circuit.recover_success_streak = circuit.recover_success_streak.saturating_add(1);
+                if circuit.recover_success_streak >= FAMILY_RECOVER_SUCCESS_STREAK_REQUIRED {
+                    circuit.fail_streak = 0;
+                    circuit.recover_success_streak = 0;
+                    circuit.suppressed_until = None;
+                    circuit.state = MeFamilyRuntimeState::Healthy;
+                    info!(
+                        ?family,
+                        "ME family suppression lifted after stable recovery probes"
+                    );
+                }
+            }
+        }
+        _ => {
+            if degraded {
+                circuit.fail_streak = circuit.fail_streak.saturating_add(1);
+                circuit.recover_success_streak = 0;
+                circuit.state = MeFamilyRuntimeState::Degraded;
+                if allow_suppress && circuit.fail_streak >= FAMILY_SUPPRESS_FAIL_STREAK_THRESHOLD {
+                    circuit.state = MeFamilyRuntimeState::Suppressed;
+                    let until = now + Duration::from_secs(FAMILY_SUPPRESS_WINDOW_SECS);
+                    circuit.suppressed_until = Some(until);
+                    circuit.next_probe_at =
+                        now + Duration::from_secs(FAMILY_RECOVER_PROBE_INTERVAL_SECS);
+                    warn!(
+                        ?family,
+                        fail_streak = circuit.fail_streak,
+                        suppress_secs = FAMILY_SUPPRESS_WINDOW_SECS,
+                        "ME family temporarily suppressed after repeated degradation"
+                    );
+                }
+            } else {
+                circuit.fail_streak = 0;
+                circuit.recover_success_streak = 0;
+                circuit.suppressed_until = None;
+                circuit.state = MeFamilyRuntimeState::Healthy;
+            }
+        }
+    }
+
+    if previous_state != circuit.state {
+        circuit.state_since_at = now;
+    }
+
+    let suppressed_until_epoch_secs = circuit
+        .suppressed_until
+        .and_then(|until| {
+            if until > now {
+                Some(
+                    now_epoch_secs
+                        .saturating_add(until.saturating_duration_since(now).as_secs()),
+                )
+            } else {
+                None
+            }
+        })
+        .unwrap_or(0);
+    let state_since_epoch_secs = if previous_state == circuit.state {
+        pool.family_runtime_state_since_epoch_secs(family)
+    } else {
+        now_epoch_secs
+    };
+    pool.set_family_runtime_state(
+        family,
+        circuit.state,
+        state_since_epoch_secs,
+        suppressed_until_epoch_secs,
+        circuit.fail_streak,
+        circuit.recover_success_streak,
+    );
+
+    !matches!(circuit.state, MeFamilyRuntimeState::Suppressed) && degraded
+}
+
 fn draining_writer_timeout_expired(
     pool: &MePool,
     writer: &MeWriter,
@@ -358,7 +577,8 @@ pub(super) async fn reap_draining_writers(
             continue;
         }
         pool.stats.increment_pool_force_close_total();
-        pool.remove_writer_and_close_clients(writer_id).await;
+        pool.remove_writer_and_close_clients(writer_id, MeWriterTeardownReason::ReapTimeoutExpired)
+            .await;
         pool.stats
             .increment_me_draining_writers_reap_progress_total();
     }
@@ -376,7 +596,8 @@ pub(super) async fn reap_draining_writers(
             continue;
         }
         pool.stats.increment_pool_force_close_total();
-        pool.remove_writer_and_close_clients(writer_id).await;
+        pool.remove_writer_and_close_clients(writer_id, MeWriterTeardownReason::ReapThresholdForce)
+            .await;
         pool.stats
             .increment_me_draining_writers_reap_progress_total();
         closed_total = closed_total.saturating_add(1);
@@ -388,7 +609,8 @@ pub(super) async fn reap_draining_writers(
         if !closed_writer_ids.insert(writer_id) {
             continue;
         }
-        pool.remove_writer_and_close_clients(writer_id).await;
+        pool.remove_writer_and_close_clients(writer_id, MeWriterTeardownReason::ReapEmpty)
+            .await;
         pool.stats
             .increment_me_draining_writers_reap_progress_total();
         closed_total = closed_total.saturating_add(1);
@@ -1550,6 +1772,187 @@ async fn maybe_rotate_single_endpoint_shadow(
     );
 }
 
+/// Last-resort safety net for draining writers stuck past their deadline.
+///
+/// Runs every `TICK_SECS` and force-closes any draining writer whose
+/// `drain_deadline_epoch_secs` has been exceeded by more than a threshold.
+///
+/// Two thresholds:
+///   - `SOFT_THRESHOLD_SECS` (60s): writers with no bound clients
+///   - `HARD_THRESHOLD_SECS` (300s): writers WITH bound clients (unconditional)
+///
+/// Intentionally kept trivial and independent of pool config to minimise
+/// the probability of panicking itself. Uses `SystemTime` directly
+/// as a fallback clock source and timeouts on every lock acquisition
+/// and writer removal so one stuck writer cannot block the rest.
+pub async fn me_zombie_writer_watchdog(pool: Arc<MePool>) {
+    use std::time::{SystemTime, UNIX_EPOCH};
+
+    const TICK_SECS: u64 = 30;
+    const SOFT_THRESHOLD_SECS: u64 = 60;
+    const HARD_THRESHOLD_SECS: u64 = 300;
+    const LOCK_TIMEOUT_SECS: u64 = 5;
+    const REMOVE_TIMEOUT_SECS: u64 = 10;
+    const HARD_DETACH_TIMEOUT_STREAK: u8 = 3;
+
+    let mut removal_timeout_streak = HashMap::<u64, u8>::new();
+
+    loop {
+        tokio::time::sleep(Duration::from_secs(TICK_SECS)).await;
+
+        let now = match SystemTime::now().duration_since(UNIX_EPOCH) {
+            Ok(d) => d.as_secs(),
+            Err(_) => continue,
+        };
+
+        // Phase 1: collect zombie IDs under a short read-lock with timeout.
+        let zombie_ids_with_meta: Vec<(u64, bool)> = {
+            let Ok(ws) = tokio::time::timeout(
+                Duration::from_secs(LOCK_TIMEOUT_SECS),
+                pool.writers.read(),
+            )
+            .await
+            else {
+                warn!("zombie_watchdog: writers read-lock timeout, skipping tick");
+                continue;
+            };
+            ws.iter()
+                .filter(|w| w.draining.load(std::sync::atomic::Ordering::Relaxed))
+                .filter_map(|w| {
+                    let deadline = w
+                        .drain_deadline_epoch_secs
+                        .load(std::sync::atomic::Ordering::Relaxed);
+                    if deadline == 0 {
+                        return None;
+                    }
+                    let overdue = now.saturating_sub(deadline);
+                    if overdue == 0 {
+                        return None;
+                    }
+                    let started = w
+                        .draining_started_at_epoch_secs
+                        .load(std::sync::atomic::Ordering::Relaxed);
+                    let drain_age = now.saturating_sub(started);
+                    if drain_age > HARD_THRESHOLD_SECS {
+                        return Some((w.id, true));
+                    }
+                    if overdue > SOFT_THRESHOLD_SECS {
+                        return Some((w.id, false));
+                    }
+                    None
+                })
+                .collect()
+        };
+        // read lock released here
+
+        if zombie_ids_with_meta.is_empty() {
+            removal_timeout_streak.clear();
+            continue;
+        }
+
+        let mut active_zombie_ids = HashSet::<u64>::with_capacity(zombie_ids_with_meta.len());
+        for (writer_id, _) in &zombie_ids_with_meta {
+            active_zombie_ids.insert(*writer_id);
+        }
+        removal_timeout_streak.retain(|writer_id, _| active_zombie_ids.contains(writer_id));
+
+        warn!(
+            zombie_count = zombie_ids_with_meta.len(),
+            soft_threshold_secs = SOFT_THRESHOLD_SECS,
+            hard_threshold_secs = HARD_THRESHOLD_SECS,
+            "Zombie draining writers detected by watchdog, force-closing"
+        );
+
+        // Phase 2: remove each writer individually with a timeout.
+        // One stuck removal cannot block the rest.
+        for (writer_id, had_clients) in &zombie_ids_with_meta {
+            let result = tokio::time::timeout(
+                Duration::from_secs(REMOVE_TIMEOUT_SECS),
+                pool.remove_writer_and_close_clients(
+                    *writer_id,
+                    MeWriterTeardownReason::WatchdogStuckDraining,
+                ),
+            )
+            .await;
+            match result {
+                Ok(true) => {
+                    removal_timeout_streak.remove(writer_id);
+                    pool.stats.increment_pool_force_close_total();
+                    pool.stats
+                        .increment_me_draining_writers_reap_progress_total();
+                    info!(
+                        writer_id,
+                        had_clients,
+                        "Zombie writer removed by watchdog"
+                    );
+                }
+                Ok(false) => {
+                    removal_timeout_streak.remove(writer_id);
+                    debug!(
+                        writer_id,
+                        had_clients,
+                        "Zombie writer watchdog removal became no-op"
+                    );
+                }
+                Err(_) => {
+                    pool.stats.increment_me_writer_teardown_timeout_total();
+                    let streak = removal_timeout_streak
+                        .entry(*writer_id)
+                        .and_modify(|value| *value = value.saturating_add(1))
+                        .or_insert(1);
+                    warn!(
+                        writer_id,
+                        had_clients,
+                        timeout_streak = *streak,
+                        "Zombie writer removal timed out"
+                    );
+                    if *streak < HARD_DETACH_TIMEOUT_STREAK {
+                        continue;
+                    }
+                    pool.stats.increment_me_writer_teardown_escalation_total();
+
+                    let hard_detach = tokio::time::timeout(
+                        Duration::from_secs(REMOVE_TIMEOUT_SECS),
+                        pool.remove_draining_writer_hard_detach(
+                            *writer_id,
+                            MeWriterTeardownReason::WatchdogStuckDraining,
+                        ),
+                    )
+                    .await;
+                    match hard_detach {
+                        Ok(true) => {
+                            removal_timeout_streak.remove(writer_id);
+                            pool.stats.increment_pool_force_close_total();
+                            pool.stats
+                                .increment_me_draining_writers_reap_progress_total();
+                            info!(
+                                writer_id,
+                                had_clients,
+                                "Zombie writer hard-detached after repeated timeouts"
+                            );
+                        }
+                        Ok(false) => {
+                            removal_timeout_streak.remove(writer_id);
+                            debug!(
+                                writer_id,
+                                had_clients,
+                                "Zombie hard-detach skipped (writer already gone or no longer draining)"
+                            );
+                        }
+                        Err(_) => {
+                            pool.stats.increment_me_writer_teardown_timeout_total();
+                            warn!(
+                                writer_id,
+                                had_clients,
+                                "Zombie hard-detach timed out, will retry next tick"
+                            );
+                        }
+                    }
+                }
+            }
+        }
+    }
+}
 #[cfg(test)]
 mod tests {
     use std::collections::HashMap;
@@ -1561,13 +1964,19 @@ mod tests {
     use tokio::sync::mpsc;
     use tokio_util::sync::CancellationToken;
 
-    use super::reap_draining_writers;
+    use super::{
+        FamilyCircuitState, apply_family_circuit_result, reap_draining_writers,
+        should_run_family_check,
+    };
     use crate::config::{GeneralConfig, MeRouteNoWriterMode, MeSocksKdfPolicy, MeWriterPickMode};
     use crate::crypto::SecureRandom;
+    use crate::network::IpFamily;
     use crate::network::probe::NetworkDecision;
     use crate::stats::Stats;
     use crate::transport::middle_proxy::codec::WriterCommand;
-    use crate::transport::middle_proxy::pool::{MePool, MeWriter, WriterContour};
+    use crate::transport::middle_proxy::pool::{
+        MeFamilyRuntimeState, MePool, MeWriter, WriterContour,
+    };
     use crate::transport::middle_proxy::registry::ConnMeta;
 
     async fn make_pool(me_pool_drain_threshold: u64) -> Arc<MePool> {
@@ -1745,4 +2154,47 @@ mod tests {
         assert_eq!(pool.registry.get_writer(conn_b).await.unwrap().writer_id, 20);
         assert_eq!(pool.registry.get_writer(conn_c).await.unwrap().writer_id, 30);
     }
+
+    #[tokio::test]
+    async fn suppressed_family_probe_skip_preserves_suppressed_state() {
+        let pool = make_pool(0).await;
+        let now = Instant::now();
+        let now_epoch_secs = MePool::now_epoch_secs();
+        let suppressed_until_epoch_secs = now_epoch_secs.saturating_add(60);
+        pool.set_family_runtime_state(
+            IpFamily::V6,
+            MeFamilyRuntimeState::Suppressed,
+            now_epoch_secs,
+            suppressed_until_epoch_secs,
+            7,
+            0,
+        );
+
+        let mut circuit = FamilyCircuitState {
+            state: MeFamilyRuntimeState::Suppressed,
+            state_since_at: now,
+            suppressed_until: Some(now + Duration::from_secs(60)),
+            next_probe_at: now + Duration::from_secs(5),
+            fail_streak: 7,
+            recover_success_streak: 0,
+        };
+
+        assert!(!should_run_family_check(&mut circuit, now));
+        assert!(!apply_family_circuit_result(
+            &pool,
+            IpFamily::V6,
+            &mut circuit,
+            None,
+            true,
+            now,
+            now_epoch_secs,
+        ));
+        assert_eq!(circuit.state, MeFamilyRuntimeState::Suppressed);
+        assert_eq!(circuit.fail_streak, 7);
+        assert_eq!(circuit.recover_success_streak, 0);
+        assert_eq!(
+            pool.family_runtime_state(IpFamily::V6),
+            MeFamilyRuntimeState::Suppressed,
+        );
+    }
 }

src/transport/middle_proxy/health_adversarial_tests.rs

@@ -316,7 +316,12 @@ async fn reap_draining_writers_maintains_warn_state_subset_property_under_bulk_c
 
         let ids = sorted_writer_ids(&pool).await;
         for writer_id in ids.into_iter().take(3) {
-            let _ = pool.remove_writer_and_close_clients(writer_id).await;
+            let _ = pool
+                .remove_writer_and_close_clients(
+                    writer_id,
+                    crate::stats::MeWriterTeardownReason::ReapEmpty,
+                )
+                .await;
         }
 
         reap_draining_writers(&pool, &mut warn_next_allowed, &mut soft_evict_next_allowed).await;

src/transport/middle_proxy/health_regression_tests.rs

@@ -12,7 +12,9 @@ use super::codec::WriterCommand;
 use super::health::{health_drain_close_budget, reap_draining_writers};
 use super::pool::{MePool, MeWriter, WriterContour};
 use super::registry::ConnMeta;
-use crate::config::{GeneralConfig, MeRouteNoWriterMode, MeSocksKdfPolicy, MeWriterPickMode};
+use crate::config::{
+    GeneralConfig, MeBindStaleMode, MeRouteNoWriterMode, MeSocksKdfPolicy, MeWriterPickMode,
+};
 use crate::crypto::SecureRandom;
 use crate::network::probe::NetworkDecision;
 use crate::stats::Stats;
@@ -195,7 +197,9 @@ async fn reap_draining_writers_drops_warn_state_for_removed_writer() {
     reap_draining_writers(&pool, &mut warn_next_allowed, &mut soft_evict_next_allowed).await;
     assert!(warn_next_allowed.contains_key(&7));
 
-    let _ = pool.remove_writer_and_close_clients(7).await;
+    let _ = pool
+        .remove_writer_and_close_clients(7, crate::stats::MeWriterTeardownReason::ReapEmpty)
+        .await;
     assert!(pool.registry.get_writer(conn_ids[0]).await.is_none());
 
     reap_draining_writers(&pool, &mut warn_next_allowed, &mut soft_evict_next_allowed).await;
@@ -525,7 +529,12 @@ async fn reap_draining_writers_warn_state_never_exceeds_live_draining_population
 
         let existing_writer_ids = current_writer_ids(&pool).await;
         for writer_id in existing_writer_ids.into_iter().take(4) {
-            let _ = pool.remove_writer_and_close_clients(writer_id).await;
+            let _ = pool
+                .remove_writer_and_close_clients(
+                    writer_id,
+                    crate::stats::MeWriterTeardownReason::ReapEmpty,
+                )
+                .await;
         }
         reap_draining_writers(&pool, &mut warn_next_allowed, &mut soft_evict_next_allowed).await;
         assert!(warn_next_allowed.len() <= pool.writers.read().await.len());
@@ -646,10 +655,23 @@ async fn reap_draining_writers_instadrain_removes_non_expired_writers_immediatel
 
 #[test]
 fn general_config_default_drain_threshold_remains_enabled() {
-    assert_eq!(GeneralConfig::default().me_pool_drain_threshold, 128);
+    assert_eq!(GeneralConfig::default().me_pool_drain_threshold, 32);
     assert!(GeneralConfig::default().me_pool_drain_soft_evict_enabled);
     assert_eq!(
-        GeneralConfig::default().me_pool_drain_soft_evict_per_writer,
+        GeneralConfig::default().me_pool_drain_soft_evict_grace_secs,
-        1
+        10
     );
+    assert_eq!(
+        GeneralConfig::default().me_pool_drain_soft_evict_per_writer,
+        2
+    );
+    assert_eq!(
+        GeneralConfig::default().me_pool_drain_soft_evict_budget_per_core,
+        16
+    );
+    assert_eq!(
+        GeneralConfig::default().me_pool_drain_soft_evict_cooldown_ms,
+        1000
+    );
+    assert_eq!(GeneralConfig::default().me_bind_stale_mode, MeBindStaleMode::Never);
 }

src/transport/middle_proxy/mod.rs

@@ -30,7 +30,7 @@ mod health_adversarial_tests;
 
 use bytes::Bytes;
 
-pub use health::{me_drain_timeout_enforcer, me_health_monitor};
+pub use health::{me_drain_timeout_enforcer, me_health_monitor, me_zombie_writer_watchdog};
 #[allow(unused_imports)]
 pub use ping::{run_me_ping, format_sample_line, format_me_route, MePingReport, MePingSample, MePingFamily};
 pub use pool::MePool;

src/transport/middle_proxy/ping.rs

@@ -7,6 +7,7 @@ use tokio::net::UdpSocket;
 use crate::config::{UpstreamConfig, UpstreamType};
 use crate::crypto::SecureRandom;
 use crate::error::ProxyError;
+use crate::transport::shadowsocks::sanitize_shadowsocks_url;
 use crate::transport::{UpstreamEgressInfo, UpstreamRouteKind};
 
 use super::MePool;
@@ -40,7 +41,11 @@ pub fn format_sample_line(sample: &MePingSample) -> String {
     let sign = if sample.dc >= 0 { "+" } else { "-" };
     let addr = format!("{}:{}", sample.addr.ip(), sample.addr.port());
 
-    match (sample.connect_ms, sample.handshake_ms.as_ref(), sample.error.as_ref()) {
+    match (
+        sample.connect_ms,
+        sample.handshake_ms.as_ref(),
+        sample.error.as_ref(),
+    ) {
         (Some(conn), Some(hs), None) => format!(
             "     {sign} {addr}\tPing: {:.0} ms / RPC: {:.0} ms / OK",
             conn, hs
@@ -121,6 +126,7 @@ fn route_from_egress(egress: Option<UpstreamEgressInfo>) -> Option<String> {
                 None => route,
             })
         }
+        UpstreamRouteKind::Shadowsocks => Some("shadowsocks".to_string()),
     }
 }
 
@@ -232,6 +238,9 @@ pub async fn format_me_route(
             }
             UpstreamType::Socks4 { address, .. } => format!("socks4://{address}"),
             UpstreamType::Socks5 { address, .. } => format!("socks5://{address}"),
+            UpstreamType::Shadowsocks { url, .. } => sanitize_shadowsocks_url(url)
+                .map(|address| format!("shadowsocks://{address}"))
+                .unwrap_or_else(|_| "shadowsocks://invalid".to_string()),
         };
     }
 
@@ -254,6 +263,12 @@ pub async fn format_me_route(
     if has_socks5 {
         kinds.push("socks5");
     }
+    if enabled_upstreams
+        .iter()
+        .any(|u| matches!(u.upstream_type, UpstreamType::Shadowsocks { .. }))
+    {
+        kinds.push("shadowsocks");
+    }
     format!("mixed upstreams ({})", kinds.join(", "))
 }
 
@@ -335,7 +350,10 @@ pub async fn run_me_ping(pool: &Arc<MePool>, rng: &SecureRandom) -> Vec<MePingRe
                 Ok((stream, conn_rtt, upstream_egress)) => {
                     connect_ms = Some(conn_rtt);
                     route = route_from_egress(upstream_egress);
-                    match pool.handshake_only(stream, addr, upstream_egress, rng).await {
+                    match pool
+                        .handshake_only(stream, addr, upstream_egress, rng)
+                        .await
+                    {
                         Ok(hs) => {
                             handshake_ms = Some(hs.handshake_ms);
                             // drop halves to close

src/transport/middle_proxy/pool.rs

@@ -18,6 +18,8 @@ use crate::transport::UpstreamManager;
 use super::ConnRegistry;
 use super::codec::WriterCommand;
 
+const ME_FORCE_CLOSE_SAFETY_FALLBACK_SECS: u64 = 300;
+
 #[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
 pub(super) struct RefillDcKey {
     pub dc: i32,
@@ -72,6 +74,64 @@ impl WriterContour {
     }
 }
 
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+#[repr(u8)]
+pub(crate) enum MeFamilyRuntimeState {
+    Healthy = 0,
+    Degraded = 1,
+    Suppressed = 2,
+    Recovering = 3,
+}
+
+impl MeFamilyRuntimeState {
+    pub(crate) fn from_u8(value: u8) -> Self {
+        match value {
+            1 => Self::Degraded,
+            2 => Self::Suppressed,
+            3 => Self::Recovering,
+            _ => Self::Healthy,
+        }
+    }
+
+    pub(crate) fn as_str(self) -> &'static str {
+        match self {
+            Self::Healthy => "healthy",
+            Self::Degraded => "degraded",
+            Self::Suppressed => "suppressed",
+            Self::Recovering => "recovering",
+        }
+    }
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+#[repr(u8)]
+pub(crate) enum MeDrainGateReason {
+    Open = 0,
+    CoverageQuorum = 1,
+    Redundancy = 2,
+    SuppressionActive = 3,
+}
+
+impl MeDrainGateReason {
+    pub(crate) fn from_u8(value: u8) -> Self {
+        match value {
+            1 => Self::CoverageQuorum,
+            2 => Self::Redundancy,
+            3 => Self::SuppressionActive,
+            _ => Self::Open,
+        }
+    }
+
+    pub(crate) fn as_str(self) -> &'static str {
+        match self {
+            Self::Open => "open",
+            Self::CoverageQuorum => "coverage_quorum",
+            Self::Redundancy => "redundancy",
+            Self::SuppressionActive => "suppression_active",
+        }
+    }
+}
+
 #[derive(Debug, Clone)]
 pub struct SecretSnapshot {
     pub epoch: u64,
@@ -201,6 +261,20 @@ pub struct MePool {
     pub(super) me_health_interval_ms_unhealthy: AtomicU64,
     pub(super) me_health_interval_ms_healthy: AtomicU64,
     pub(super) me_warn_rate_limit_ms: AtomicU64,
+    pub(super) me_family_v4_runtime_state: AtomicU8,
+    pub(super) me_family_v6_runtime_state: AtomicU8,
+    pub(super) me_family_v4_state_since_epoch_secs: AtomicU64,
+    pub(super) me_family_v6_state_since_epoch_secs: AtomicU64,
+    pub(super) me_family_v4_suppressed_until_epoch_secs: AtomicU64,
+    pub(super) me_family_v6_suppressed_until_epoch_secs: AtomicU64,
+    pub(super) me_family_v4_fail_streak: AtomicU32,
+    pub(super) me_family_v6_fail_streak: AtomicU32,
+    pub(super) me_family_v4_recover_success_streak: AtomicU32,
+    pub(super) me_family_v6_recover_success_streak: AtomicU32,
+    pub(super) me_last_drain_gate_route_quorum_ok: AtomicBool,
+    pub(super) me_last_drain_gate_redundancy_ok: AtomicBool,
+    pub(super) me_last_drain_gate_block_reason: AtomicU8,
+    pub(super) me_last_drain_gate_updated_at_epoch_secs: AtomicU64,
     pub(super) runtime_ready: AtomicBool,
     pool_size: usize,
     pub(super) preferred_endpoints_by_dc: Arc<RwLock<HashMap<i32, Vec<SocketAddr>>>>,
@@ -229,6 +303,14 @@ impl MePool {
             .as_secs()
     }
 
+    fn normalize_force_close_secs(force_close_secs: u64) -> u64 {
+        if force_close_secs == 0 {
+            ME_FORCE_CLOSE_SAFETY_FALLBACK_SECS
+        } else {
+            force_close_secs
+        }
+    }
+
     pub fn new(
         proxy_tag: Option<Vec<u8>>,
         proxy_secret: Vec<u8>,
@@ -477,7 +559,9 @@ impl MePool {
             me_pool_drain_soft_evict_cooldown_ms: AtomicU64::new(
                 me_pool_drain_soft_evict_cooldown_ms.max(1),
             ),
-            me_pool_force_close_secs: AtomicU64::new(me_pool_force_close_secs),
+            me_pool_force_close_secs: AtomicU64::new(Self::normalize_force_close_secs(
+                me_pool_force_close_secs,
+            )),
             me_pool_min_fresh_ratio_permille: AtomicU32::new(Self::ratio_to_permille(
                 me_pool_min_fresh_ratio,
             )),
@@ -506,6 +590,20 @@ impl MePool {
             me_health_interval_ms_unhealthy: AtomicU64::new(me_health_interval_ms_unhealthy.max(1)),
             me_health_interval_ms_healthy: AtomicU64::new(me_health_interval_ms_healthy.max(1)),
             me_warn_rate_limit_ms: AtomicU64::new(me_warn_rate_limit_ms.max(1)),
+            me_family_v4_runtime_state: AtomicU8::new(MeFamilyRuntimeState::Healthy as u8),
+            me_family_v6_runtime_state: AtomicU8::new(MeFamilyRuntimeState::Healthy as u8),
+            me_family_v4_state_since_epoch_secs: AtomicU64::new(Self::now_epoch_secs()),
+            me_family_v6_state_since_epoch_secs: AtomicU64::new(Self::now_epoch_secs()),
+            me_family_v4_suppressed_until_epoch_secs: AtomicU64::new(0),
+            me_family_v6_suppressed_until_epoch_secs: AtomicU64::new(0),
+            me_family_v4_fail_streak: AtomicU32::new(0),
+            me_family_v6_fail_streak: AtomicU32::new(0),
+            me_family_v4_recover_success_streak: AtomicU32::new(0),
+            me_family_v6_recover_success_streak: AtomicU32::new(0),
+            me_last_drain_gate_route_quorum_ok: AtomicBool::new(false),
+            me_last_drain_gate_redundancy_ok: AtomicBool::new(false),
+            me_last_drain_gate_block_reason: AtomicU8::new(MeDrainGateReason::Open as u8),
+            me_last_drain_gate_updated_at_epoch_secs: AtomicU64::new(Self::now_epoch_secs()),
             runtime_ready: AtomicBool::new(false),
             preferred_endpoints_by_dc: Arc::new(RwLock::new(preferred_endpoints_by_dc)),
         })
@@ -523,6 +621,153 @@ impl MePool {
         self.runtime_ready.load(Ordering::Relaxed)
     }
 
+    pub(super) fn set_family_runtime_state(
+        &self,
+        family: IpFamily,
+        state: MeFamilyRuntimeState,
+        state_since_epoch_secs: u64,
+        suppressed_until_epoch_secs: u64,
+        fail_streak: u32,
+        recover_success_streak: u32,
+    ) {
+        match family {
+            IpFamily::V4 => {
+                self.me_family_v4_runtime_state
+                    .store(state as u8, Ordering::Relaxed);
+                self.me_family_v4_state_since_epoch_secs
+                    .store(state_since_epoch_secs, Ordering::Relaxed);
+                self.me_family_v4_suppressed_until_epoch_secs
+                    .store(suppressed_until_epoch_secs, Ordering::Relaxed);
+                self.me_family_v4_fail_streak
+                    .store(fail_streak, Ordering::Relaxed);
+                self.me_family_v4_recover_success_streak
+                    .store(recover_success_streak, Ordering::Relaxed);
+            }
+            IpFamily::V6 => {
+                self.me_family_v6_runtime_state
+                    .store(state as u8, Ordering::Relaxed);
+                self.me_family_v6_state_since_epoch_secs
+                    .store(state_since_epoch_secs, Ordering::Relaxed);
+                self.me_family_v6_suppressed_until_epoch_secs
+                    .store(suppressed_until_epoch_secs, Ordering::Relaxed);
+                self.me_family_v6_fail_streak
+                    .store(fail_streak, Ordering::Relaxed);
+                self.me_family_v6_recover_success_streak
+                    .store(recover_success_streak, Ordering::Relaxed);
+            }
+        }
+    }
+
+    pub(crate) fn family_runtime_state(&self, family: IpFamily) -> MeFamilyRuntimeState {
+        match family {
+            IpFamily::V4 => MeFamilyRuntimeState::from_u8(
+                self.me_family_v4_runtime_state.load(Ordering::Relaxed),
+            ),
+            IpFamily::V6 => MeFamilyRuntimeState::from_u8(
+                self.me_family_v6_runtime_state.load(Ordering::Relaxed),
+            ),
+        }
+    }
+
+    pub(crate) fn family_runtime_state_since_epoch_secs(&self, family: IpFamily) -> u64 {
+        match family {
+            IpFamily::V4 => self
+                .me_family_v4_state_since_epoch_secs
+                .load(Ordering::Relaxed),
+            IpFamily::V6 => self
+                .me_family_v6_state_since_epoch_secs
+                .load(Ordering::Relaxed),
+        }
+    }
+
+    pub(crate) fn family_suppressed_until_epoch_secs(&self, family: IpFamily) -> u64 {
+        match family {
+            IpFamily::V4 => self
+                .me_family_v4_suppressed_until_epoch_secs
+                .load(Ordering::Relaxed),
+            IpFamily::V6 => self
+                .me_family_v6_suppressed_until_epoch_secs
+                .load(Ordering::Relaxed),
+        }
+    }
+
+    pub(crate) fn family_fail_streak(&self, family: IpFamily) -> u32 {
+        match family {
+            IpFamily::V4 => self.me_family_v4_fail_streak.load(Ordering::Relaxed),
+            IpFamily::V6 => self.me_family_v6_fail_streak.load(Ordering::Relaxed),
+        }
+    }
+
+    pub(crate) fn family_recover_success_streak(&self, family: IpFamily) -> u32 {
+        match family {
+            IpFamily::V4 => self
+                .me_family_v4_recover_success_streak
+                .load(Ordering::Relaxed),
+            IpFamily::V6 => self
+                .me_family_v6_recover_success_streak
+                .load(Ordering::Relaxed),
+        }
+    }
+
+    pub(crate) fn is_family_temporarily_suppressed(
+        &self,
+        family: IpFamily,
+        now_epoch_secs: u64,
+    ) -> bool {
+        self.family_suppressed_until_epoch_secs(family) > now_epoch_secs
+    }
+
+    pub(super) fn family_enabled_for_drain_coverage(
+        &self,
+        family: IpFamily,
+        now_epoch_secs: u64,
+    ) -> bool {
+        let configured = match family {
+            IpFamily::V4 => self.decision.ipv4_me,
+            IpFamily::V6 => self.decision.ipv6_me,
+        };
+        configured && !self.is_family_temporarily_suppressed(family, now_epoch_secs)
+    }
+
+    pub(super) fn set_last_drain_gate(
+        &self,
+        route_quorum_ok: bool,
+        redundancy_ok: bool,
+        block_reason: MeDrainGateReason,
+        updated_at_epoch_secs: u64,
+    ) {
+        self.me_last_drain_gate_route_quorum_ok
+            .store(route_quorum_ok, Ordering::Relaxed);
+        self.me_last_drain_gate_redundancy_ok
+            .store(redundancy_ok, Ordering::Relaxed);
+        self.me_last_drain_gate_block_reason
+            .store(block_reason as u8, Ordering::Relaxed);
+        self.me_last_drain_gate_updated_at_epoch_secs
+            .store(updated_at_epoch_secs, Ordering::Relaxed);
+    }
+
+    pub(crate) fn last_drain_gate_route_quorum_ok(&self) -> bool {
+        self.me_last_drain_gate_route_quorum_ok
+            .load(Ordering::Relaxed)
+    }
+
+    pub(crate) fn last_drain_gate_redundancy_ok(&self) -> bool {
+        self.me_last_drain_gate_redundancy_ok
+            .load(Ordering::Relaxed)
+    }
+
+    pub(crate) fn last_drain_gate_block_reason(&self) -> MeDrainGateReason {
+        MeDrainGateReason::from_u8(
+            self.me_last_drain_gate_block_reason
+                .load(Ordering::Relaxed),
+        )
+    }
+
+    pub(crate) fn last_drain_gate_updated_at_epoch_secs(&self) -> u64 {
+        self.me_last_drain_gate_updated_at_epoch_secs
+            .load(Ordering::Relaxed)
+    }
+
     pub fn update_runtime_reinit_policy(
         &self,
         hardswap: bool,
@@ -587,8 +832,10 @@ impl MePool {
         );
         self.me_pool_drain_soft_evict_cooldown_ms
             .store(pool_drain_soft_evict_cooldown_ms.max(1), Ordering::Relaxed);
-        self.me_pool_force_close_secs
+        self.me_pool_force_close_secs.store(
-            .store(force_close_secs, Ordering::Relaxed);
+            Self::normalize_force_close_secs(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
@@ -733,12 +980,9 @@ impl MePool {
     }
 
     pub(super) fn force_close_timeout(&self) -> Option<Duration> {
-        let secs = self.me_pool_force_close_secs.load(Ordering::Relaxed);
+        let secs =
-        if secs == 0 {
+            Self::normalize_force_close_secs(self.me_pool_force_close_secs.load(Ordering::Relaxed));
-            None
+        Some(Duration::from_secs(secs))
-        } else {
-            Some(Duration::from_secs(secs))
-        }
     }
 
     pub(super) fn drain_soft_evict_enabled(&self) -> bool {
@@ -1010,9 +1254,10 @@ impl MePool {
     }
 
     pub(super) async fn active_coverage_required_total(&self) -> usize {
+        let now_epoch_secs = Self::now_epoch_secs();
         let mut endpoints_by_dc = HashMap::<i32, HashSet<SocketAddr>>::new();
 
-        if self.decision.ipv4_me {
+        if self.family_enabled_for_drain_coverage(IpFamily::V4, now_epoch_secs) {
             let map = self.proxy_map_v4.read().await;
             for (dc, addrs) in map.iter() {
                 let entry = endpoints_by_dc.entry(*dc).or_default();
@@ -1022,7 +1267,7 @@ impl MePool {
             }
         }
 
-        if self.decision.ipv6_me {
+        if self.family_enabled_for_drain_coverage(IpFamily::V6, now_epoch_secs) {
             let map = self.proxy_map_v6.read().await;
             for (dc, addrs) in map.iter() {
                 let entry = endpoints_by_dc.entry(*dc).or_default();

src/transport/middle_proxy/pool_refill.rs

@@ -74,9 +74,8 @@ impl MePool {
             debug!(
                 %addr,
                 wait_ms = expiry.saturating_duration_since(now).as_millis(),
-                "All ME endpoints are quarantined for the DC group; retrying earliest one"
+                "All ME endpoints are quarantined for the DC group; waiting for quarantine expiry"
             );
-            return vec![addr];
         }
 
         Vec::new()
@@ -165,9 +164,10 @@ impl MePool {
     }
 
     async fn endpoints_for_dc(&self, target_dc: i32) -> Vec<SocketAddr> {
+        let now_epoch_secs = Self::now_epoch_secs();
         let mut endpoints = HashSet::<SocketAddr>::new();
 
-        if self.decision.ipv4_me {
+        if self.family_enabled_for_drain_coverage(IpFamily::V4, now_epoch_secs) {
             let map = self.proxy_map_v4.read().await;
             if let Some(addrs) = map.get(&target_dc) {
                 for (ip, port) in addrs {
@@ -176,7 +176,7 @@ impl MePool {
             }
         }
 
-        if self.decision.ipv6_me {
+        if self.family_enabled_for_drain_coverage(IpFamily::V6, now_epoch_secs) {
             let map = self.proxy_map_v6.read().await;
             if let Some(addrs) = map.get(&target_dc) {
                 for (ip, port) in addrs {

src/transport/middle_proxy/pool_reinit.rs

@@ -11,8 +11,9 @@ use tracing::{debug, info, warn};
 use std::collections::hash_map::DefaultHasher;
 
 use crate::crypto::SecureRandom;
+use crate::network::IpFamily;
 
-use super::pool::{MePool, WriterContour};
+use super::pool::{MeDrainGateReason, MePool, WriterContour};
 
 const ME_HARDSWAP_PENDING_TTL_SECS: u64 = 1800;
 
@@ -120,9 +121,10 @@ impl MePool {
     }
 
     async fn desired_dc_endpoints(&self) -> HashMap<i32, HashSet<SocketAddr>> {
+        let now_epoch_secs = Self::now_epoch_secs();
         let mut out: HashMap<i32, HashSet<SocketAddr>> = HashMap::new();
 
-        if self.decision.ipv4_me {
+        if self.family_enabled_for_drain_coverage(IpFamily::V4, now_epoch_secs) {
             let map_v4 = self.proxy_map_v4.read().await.clone();
             for (dc, addrs) in map_v4 {
                 let entry = out.entry(dc).or_default();
@@ -132,7 +134,7 @@ impl MePool {
             }
         }
 
-        if self.decision.ipv6_me {
+        if self.family_enabled_for_drain_coverage(IpFamily::V6, now_epoch_secs) {
             let map_v6 = self.proxy_map_v6.read().await.clone();
             for (dc, addrs) in map_v6 {
                 let entry = out.entry(dc).or_default();
@@ -313,13 +315,23 @@ impl MePool {
 
     pub async fn zero_downtime_reinit_after_map_change(self: &Arc<Self>, rng: &SecureRandom) {
         let desired_by_dc = self.desired_dc_endpoints().await;
+        let now_epoch_secs = Self::now_epoch_secs();
+        let v4_suppressed = self.is_family_temporarily_suppressed(IpFamily::V4, now_epoch_secs);
+        let v6_suppressed = self.is_family_temporarily_suppressed(IpFamily::V6, now_epoch_secs);
         if desired_by_dc.is_empty() {
             warn!("ME endpoint map is empty; skipping stale writer drain");
+            let reason = if (self.decision.ipv4_me && v4_suppressed)
+                || (self.decision.ipv6_me && v6_suppressed)
+            {
+                MeDrainGateReason::SuppressionActive
+            } else {
+                MeDrainGateReason::CoverageQuorum
+            };
+            self.set_last_drain_gate(false, false, reason, now_epoch_secs);
             return;
         }
 
         let desired_map_hash = Self::desired_map_hash(&desired_by_dc);
-        let now_epoch_secs = Self::now_epoch_secs();
         let previous_generation = self.current_generation();
         let hardswap = self.hardswap.load(Ordering::Relaxed);
         let generation = if hardswap {
@@ -390,7 +402,17 @@ impl MePool {
                 .load(Ordering::Relaxed),
         );
         let (coverage_ratio, missing_dc) = Self::coverage_ratio(&desired_by_dc, &active_writer_addrs);
+        let mut route_quorum_ok = coverage_ratio >= min_ratio;
+        let mut redundancy_ok = missing_dc.is_empty();
+        let mut redundancy_missing_dc = missing_dc.clone();
+        let mut gate_coverage_ratio = coverage_ratio;
         if !hardswap && coverage_ratio < min_ratio {
+            self.set_last_drain_gate(
+                false,
+                redundancy_ok,
+                MeDrainGateReason::CoverageQuorum,
+                now_epoch_secs,
+            );
             warn!(
                 previous_generation,
                 generation,
@@ -411,7 +433,17 @@ impl MePool {
                 .collect();
             let (fresh_coverage_ratio, fresh_missing_dc) =
                 Self::coverage_ratio(&desired_by_dc, &fresh_writer_addrs);
-            if !fresh_missing_dc.is_empty() {
+            route_quorum_ok = fresh_coverage_ratio >= min_ratio;
+            redundancy_ok = fresh_missing_dc.is_empty();
+            redundancy_missing_dc = fresh_missing_dc.clone();
+            gate_coverage_ratio = fresh_coverage_ratio;
+            if fresh_coverage_ratio < min_ratio {
+                self.set_last_drain_gate(
+                    false,
+                    redundancy_ok,
+                    MeDrainGateReason::CoverageQuorum,
+                    now_epoch_secs,
+                );
                 warn!(
                     previous_generation,
                     generation,
@@ -421,13 +453,16 @@ impl MePool {
                 );
                 return;
             }
-        } else if !missing_dc.is_empty() {
+        }
+
+        self.set_last_drain_gate(route_quorum_ok, redundancy_ok, MeDrainGateReason::Open, now_epoch_secs);
+        if !redundancy_ok {
             warn!(
-                missing_dc = ?missing_dc,
+                missing_dc = ?redundancy_missing_dc,
-                // Keep stale writers alive when fresh coverage is incomplete.
+                coverage_ratio = format_args!("{gate_coverage_ratio:.3}"),
-                "ME reinit coverage incomplete; keeping stale writers"
+                min_ratio = format_args!("{min_ratio:.3}"),
+                "ME reinit proceeds with weighted quorum while some DC groups remain uncovered"
             );
-            return;
         }
 
         if hardswap {

src/transport/middle_proxy/pool_runtime_api.rs

@@ -1,7 +1,7 @@
 use std::collections::HashMap;
 use std::time::Instant;
 
-use super::pool::{MePool, RefillDcKey};
+use super::pool::{MeDrainGateReason, MePool, RefillDcKey};
 use crate::network::IpFamily;
 
 #[derive(Clone, Debug)]
@@ -36,6 +36,24 @@ pub(crate) struct MeApiNatStunSnapshot {
     pub stun_backoff_remaining_ms: Option<u64>,
 }
 
+#[derive(Clone, Debug)]
+pub(crate) struct MeApiFamilyStateSnapshot {
+    pub family: &'static str,
+    pub state: &'static str,
+    pub state_since_epoch_secs: u64,
+    pub suppressed_until_epoch_secs: Option<u64>,
+    pub fail_streak: u32,
+    pub recover_success_streak: u32,
+}
+
+#[derive(Clone, Debug)]
+pub(crate) struct MeApiDrainGateSnapshot {
+    pub route_quorum_ok: bool,
+    pub redundancy_ok: bool,
+    pub block_reason: &'static str,
+    pub updated_at_epoch_secs: u64,
+}
+
 impl MePool {
     pub(crate) async fn api_refill_snapshot(&self) -> MeApiRefillSnapshot {
         let inflight_endpoints_total = self.refill_inflight.lock().await.len();
@@ -125,4 +143,35 @@ impl MePool {
             stun_backoff_remaining_ms,
         }
     }
+
+    pub(crate) fn api_family_state_snapshot(&self) -> Vec<MeApiFamilyStateSnapshot> {
+        [IpFamily::V4, IpFamily::V6]
+            .into_iter()
+            .map(|family| {
+                let state = self.family_runtime_state(family);
+                let suppressed_until = self.family_suppressed_until_epoch_secs(family);
+                MeApiFamilyStateSnapshot {
+                    family: match family {
+                        IpFamily::V4 => "v4",
+                        IpFamily::V6 => "v6",
+                    },
+                    state: state.as_str(),
+                    state_since_epoch_secs: self.family_runtime_state_since_epoch_secs(family),
+                    suppressed_until_epoch_secs: (suppressed_until != 0).then_some(suppressed_until),
+                    fail_streak: self.family_fail_streak(family),
+                    recover_success_streak: self.family_recover_success_streak(family),
+                }
+            })
+            .collect()
+    }
+
+    pub(crate) fn api_drain_gate_snapshot(&self) -> MeApiDrainGateSnapshot {
+        let reason: MeDrainGateReason = self.last_drain_gate_block_reason();
+        MeApiDrainGateSnapshot {
+            route_quorum_ok: self.last_drain_gate_route_quorum_ok(),
+            redundancy_ok: self.last_drain_gate_redundancy_ok(),
+            block_reason: reason.as_str(),
+            updated_at_epoch_secs: self.last_drain_gate_updated_at_epoch_secs(),
+        }
+    }
 }

src/transport/middle_proxy/pool_writer.rs

@@ -16,11 +16,13 @@ use crate::config::MeBindStaleMode;
 use crate::crypto::SecureRandom;
 use crate::error::{ProxyError, Result};
 use crate::protocol::constants::{RPC_CLOSE_EXT_U32, RPC_PING_U32};
+use crate::stats::{
+    MeWriterCleanupSideEffectStep, MeWriterTeardownMode, MeWriterTeardownReason,
+};
 
 use super::codec::{RpcWriter, WriterCommand};
 use super::pool::{MePool, MeWriter, WriterContour};
 use super::reader::reader_loop;
-use super::registry::BoundConn;
 use super::wire::build_proxy_req_payload;
 
 const ME_ACTIVE_PING_SECS: u64 = 25;
@@ -28,6 +30,12 @@ const ME_ACTIVE_PING_JITTER_SECS: i64 = 5;
 const ME_IDLE_KEEPALIVE_MAX_SECS: u64 = 5;
 const ME_RPC_PROXY_REQ_RESPONSE_WAIT_MS: u64 = 700;
 
+#[derive(Clone, Copy)]
+enum WriterRemoveGuardMode {
+    Any,
+    DrainingOnly,
+}
+
 fn is_me_peer_closed_error(error: &ProxyError) -> bool {
     matches!(error, ProxyError::Io(ioe) if ioe.kind() == ErrorKind::UnexpectedEof)
 }
@@ -44,9 +52,16 @@ impl MePool {
 
         for writer_id in closed_writer_ids {
             if self.registry.is_writer_empty(writer_id).await {
-                let _ = self.remove_writer_only(writer_id).await;
+                let _ = self
+                    .remove_writer_only(writer_id, MeWriterTeardownReason::PruneClosedWriter)
+                    .await;
             } else {
-                let _ = self.remove_writer_and_close_clients(writer_id).await;
+                let _ = self
+                    .remove_writer_and_close_clients(
+                        writer_id,
+                        MeWriterTeardownReason::PruneClosedWriter,
+                    )
+                    .await;
             }
         }
     }
@@ -143,6 +158,9 @@ impl MePool {
             crc_mode: hs.crc_mode,
         };
         let cancel_wr = cancel.clone();
+        let cleanup_done = Arc::new(AtomicBool::new(false));
+        let cleanup_for_writer = cleanup_done.clone();
+        let pool_writer_task = Arc::downgrade(self);
         tokio::spawn(async move {
             loop {
                 tokio::select! {
@@ -160,6 +178,20 @@ impl MePool {
                     _ = cancel_wr.cancelled() => break,
                 }
             }
+            if cleanup_for_writer
+                .compare_exchange(false, true, Ordering::AcqRel, Ordering::Relaxed)
+                .is_ok()
+            {
+                if let Some(pool) = pool_writer_task.upgrade() {
+                    pool.remove_writer_and_close_clients(
+                        writer_id,
+                        MeWriterTeardownReason::WriterTaskExit,
+                    )
+                    .await;
+                } else {
+                    cancel_wr.cancel();
+                }
+            }
         });
         let writer = MeWriter {
             id: writer_id,
@@ -196,7 +228,6 @@ impl MePool {
         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;
@@ -242,21 +273,29 @@ impl MePool {
                 stats_reader_close.increment_me_idle_close_by_peer_total();
                 info!(writer_id, "ME socket closed by peer on idle writer");
             }
-            if let Some(pool) = pool.upgrade()
+            if cleanup_for_reader
-                && cleanup_for_reader
+                .compare_exchange(false, true, Ordering::AcqRel, Ordering::Relaxed)
-                    .compare_exchange(false, true, Ordering::AcqRel, Ordering::Relaxed)
+                .is_ok()
-                    .is_ok()
             {
-                pool.remove_writer_and_close_clients(writer_id).await;
+                if let Some(pool) = pool.upgrade() {
+                    pool.remove_writer_and_close_clients(
+                        writer_id,
+                        MeWriterTeardownReason::ReaderExit,
+                    )
+                    .await;
+                } else {
+                    // Fallback for shutdown races: make writer task exit quickly so stale
+                    // channels are observable by periodic prune.
+                    cancel_reader_token.cancel();
+                }
             }
             if let Err(e) = res {
                 if !idle_close_by_peer {
                     warn!(error = %e, "ME reader ended");
                 }
             }
-            let mut ws = writers_arc.write().await;
+            let remaining = writers_arc.read().await.len();
-            ws.retain(|w| w.id != writer_id);
+            debug!(writer_id, remaining, "ME reader task finished");
-            info!(remaining = ws.len(), "Dead ME writer removed from pool");
         });
 
         let pool_ping = Arc::downgrade(self);
@@ -351,7 +390,11 @@ impl MePool {
                             .compare_exchange(false, true, Ordering::AcqRel, Ordering::Relaxed)
                             .is_ok()
                     {
-                        pool.remove_writer_and_close_clients(writer_id).await;
+                        pool.remove_writer_and_close_clients(
+                            writer_id,
+                            MeWriterTeardownReason::PingSendFail,
+                        )
+                        .await;
                     }
                     break;
                 }
@@ -444,7 +487,11 @@ impl MePool {
                         .compare_exchange(false, true, Ordering::AcqRel, Ordering::Relaxed)
                         .is_ok()
                     {
-                        pool.remove_writer_and_close_clients(writer_id).await;
+                        pool.remove_writer_and_close_clients(
+                            writer_id,
+                            MeWriterTeardownReason::SignalSendFail,
+                        )
+                        .await;
                     }
                     break;
                 }
@@ -478,7 +525,11 @@ impl MePool {
                         .compare_exchange(false, true, Ordering::AcqRel, Ordering::Relaxed)
                         .is_ok()
                     {
-                        pool.remove_writer_and_close_clients(writer_id).await;
+                        pool.remove_writer_and_close_clients(
+                            writer_id,
+                            MeWriterTeardownReason::SignalSendFail,
+                        )
+                        .await;
                     }
                     break;
                 }
@@ -491,21 +542,83 @@ impl MePool {
         Ok(())
     }
 
-    pub(crate) async fn remove_writer_and_close_clients(self: &Arc<Self>, writer_id: u64) {
+    pub(crate) async fn remove_writer_and_close_clients(
+        self: &Arc<Self>,
+        writer_id: u64,
+        reason: MeWriterTeardownReason,
+    ) -> bool {
         // Full client cleanup now happens inside `registry.writer_lost` to keep
         // writer reap/remove paths strictly non-blocking per connection.
-        let _ = self.remove_writer_only(writer_id).await;
+        self.remove_writer_with_mode(
+            writer_id,
+            reason,
+            MeWriterTeardownMode::Normal,
+            WriterRemoveGuardMode::Any,
+        )
+        .await
     }
 
-    async fn remove_writer_only(self: &Arc<Self>, writer_id: u64) -> Vec<BoundConn> {
+    pub(super) async fn remove_draining_writer_hard_detach(
+        self: &Arc<Self>,
+        writer_id: u64,
+        reason: MeWriterTeardownReason,
+    ) -> bool {
+        self.remove_writer_with_mode(
+            writer_id,
+            reason,
+            MeWriterTeardownMode::HardDetach,
+            WriterRemoveGuardMode::DrainingOnly,
+        )
+        .await
+    }
+
+    async fn remove_writer_only(
+        self: &Arc<Self>,
+        writer_id: u64,
+        reason: MeWriterTeardownReason,
+    ) -> bool {
+        self.remove_writer_with_mode(
+            writer_id,
+            reason,
+            MeWriterTeardownMode::Normal,
+            WriterRemoveGuardMode::Any,
+        )
+        .await
+    }
+
+    // Authoritative teardown primitive shared by normal cleanup and watchdog path.
+    // Lock-order invariant:
+    // 1) mutate `writers` under pool write lock,
+    // 2) release pool lock,
+    // 3) run registry/metrics/refill side effects.
+    // `registry.writer_lost` must never run while `writers` lock is held.
+    async fn remove_writer_with_mode(
+        self: &Arc<Self>,
+        writer_id: u64,
+        reason: MeWriterTeardownReason,
+        mode: MeWriterTeardownMode,
+        guard_mode: WriterRemoveGuardMode,
+    ) -> bool {
+        let started_at = Instant::now();
+        self.stats
+            .increment_me_writer_teardown_attempt_total(reason, mode);
         let mut close_tx: Option<mpsc::Sender<WriterCommand>> = None;
         let mut removed_addr: Option<SocketAddr> = None;
         let mut removed_dc: Option<i32> = None;
         let mut removed_uptime: Option<Duration> = None;
         let mut trigger_refill = false;
+        let mut removed = false;
         {
             let mut ws = self.writers.write().await;
             if let Some(pos) = ws.iter().position(|w| w.id == writer_id) {
+                if matches!(guard_mode, WriterRemoveGuardMode::DrainingOnly)
+                    && !ws[pos].draining.load(Ordering::Relaxed)
+                {
+                    self.stats.increment_me_writer_teardown_noop_total();
+                    self.stats
+                        .observe_me_writer_teardown_duration(mode, started_at.elapsed());
+                    return false;
+                }
                 let w = ws.remove(pos);
                 let was_draining = w.draining.load(Ordering::Relaxed);
                 if was_draining {
@@ -522,6 +635,7 @@ impl MePool {
                 }
                 close_tx = Some(w.tx.clone());
                 self.conn_count.fetch_sub(1, Ordering::Relaxed);
+                removed = true;
             }
         }
         // State invariant:
@@ -529,7 +643,7 @@ impl MePool {
         // - writer is removed from registry routing/binding maps via `writer_lost`.
         // The close command below is only a best-effort accelerator for task shutdown.
         // Cleanup progress must never depend on command-channel availability.
-        let conns = self.registry.writer_lost(writer_id).await;
+        let _ = self.registry.writer_lost(writer_id).await;
         {
             let mut tracker = self.ping_tracker.lock().await;
             tracker.retain(|_, (_, wid)| *wid != writer_id);
@@ -542,6 +656,9 @@ impl MePool {
                     self.stats.increment_me_writer_close_signal_drop_total();
                     self.stats
                         .increment_me_writer_close_signal_channel_full_total();
+                    self.stats.increment_me_writer_cleanup_side_effect_failures_total(
+                        MeWriterCleanupSideEffectStep::CloseSignalChannelFull,
+                    );
                     debug!(
                         writer_id,
                         "Skipping close signal for removed writer: command channel is full"
@@ -549,6 +666,9 @@ impl MePool {
                 }
                 Err(TrySendError::Closed(_)) => {
                     self.stats.increment_me_writer_close_signal_drop_total();
+                    self.stats.increment_me_writer_cleanup_side_effect_failures_total(
+                        MeWriterCleanupSideEffectStep::CloseSignalChannelClosed,
+                    );
                     debug!(
                         writer_id,
                         "Skipping close signal for removed writer: command channel is closed"
@@ -556,16 +676,24 @@ impl MePool {
                 }
             }
         }
-        if trigger_refill
+        if let Some(addr) = removed_addr {
-            && let Some(addr) = removed_addr
-            && let Some(writer_dc) = removed_dc
-        {
             if let Some(uptime) = removed_uptime {
                 self.maybe_quarantine_flapping_endpoint(addr, uptime).await;
             }
-            self.trigger_immediate_refill_for_dc(addr, writer_dc);
+            if trigger_refill
+                && let Some(writer_dc) = removed_dc
+            {
+                self.trigger_immediate_refill_for_dc(addr, writer_dc);
+            }
         }
-        conns
+        if removed {
+            self.stats.increment_me_writer_teardown_success_total(mode);
+        } else {
+            self.stats.increment_me_writer_teardown_noop_total();
+        }
+        self.stats
+            .observe_me_writer_teardown_duration(mode, started_at.elapsed());
+        removed
     }
 
     pub(crate) async fn mark_writer_draining_with_timeout(

src/transport/middle_proxy/send.rs

@@ -14,6 +14,7 @@ use crate::config::{MeRouteNoWriterMode, MeWriterPickMode};
 use crate::error::{ProxyError, Result};
 use crate::network::IpFamily;
 use crate::protocol::constants::{RPC_CLOSE_CONN_U32, RPC_CLOSE_EXT_U32};
+use crate::stats::MeWriterTeardownReason;
 
 use super::MePool;
 use super::codec::WriterCommand;
@@ -134,7 +135,11 @@ impl MePool {
                             Ok(()) => return Ok(()),
                             Err(TimedSendError::Closed(_)) => {
                                 warn!(writer_id = current.writer_id, "ME writer channel closed");
-                                self.remove_writer_and_close_clients(current.writer_id).await;
+                                self.remove_writer_and_close_clients(
+                                    current.writer_id,
+                                    MeWriterTeardownReason::RouteChannelClosed,
+                                )
+                                .await;
                                 continue;
                             }
                             Err(TimedSendError::Timeout(_)) => {
@@ -151,7 +156,11 @@ impl MePool {
                     }
                     Err(TrySendError::Closed(_)) => {
                         warn!(writer_id = current.writer_id, "ME writer channel closed");
-                        self.remove_writer_and_close_clients(current.writer_id).await;
+                        self.remove_writer_and_close_clients(
+                            current.writer_id,
+                            MeWriterTeardownReason::RouteChannelClosed,
+                        )
+                        .await;
                         continue;
                     }
                 }
@@ -458,7 +467,11 @@ impl MePool {
                     Err(TrySendError::Closed(_)) => {
                         self.stats.increment_me_writer_pick_closed_total(pick_mode);
                         warn!(writer_id = w.id, "ME writer channel closed");
-                        self.remove_writer_and_close_clients(w.id).await;
+                        self.remove_writer_and_close_clients(
+                            w.id,
+                            MeWriterTeardownReason::RouteChannelClosed,
+                        )
+                        .await;
                         continue;
                     }
                 }
@@ -503,7 +516,11 @@ impl MePool {
                 Err(TimedSendError::Closed(_)) => {
                     self.stats.increment_me_writer_pick_closed_total(pick_mode);
                     warn!(writer_id = w.id, "ME writer channel closed (blocking)");
-                    self.remove_writer_and_close_clients(w.id).await;
+                    self.remove_writer_and_close_clients(
+                        w.id,
+                        MeWriterTeardownReason::RouteChannelClosed,
+                    )
+                    .await;
                 }
                 Err(TimedSendError::Timeout(_)) => {
                     self.stats.increment_me_writer_pick_full_total(pick_mode);
@@ -654,7 +671,11 @@ impl MePool {
                 }
                 Err(TrySendError::Closed(_)) => {
                     debug!("ME close write failed");
-                    self.remove_writer_and_close_clients(w.writer_id).await;
+                    self.remove_writer_and_close_clients(
+                        w.writer_id,
+                        MeWriterTeardownReason::CloseRpcChannelClosed,
+                    )
+                    .await;
                 }
             }
         } else {

src/transport/mod.rs

@@ -2,6 +2,7 @@
 
 pub mod pool;
 pub mod proxy_protocol;
+pub mod shadowsocks;
 pub mod socket;
 pub mod socks;
 pub mod upstream;
@@ -14,5 +15,8 @@ pub use socket::*;
 #[allow(unused_imports)]
 pub use socks::*;
 #[allow(unused_imports)]
-pub use upstream::{DcPingResult, StartupPingResult, UpstreamEgressInfo, UpstreamManager, UpstreamRouteKind};
+pub use upstream::{
+    DcPingResult, StartupPingResult, UpstreamEgressInfo, UpstreamManager, UpstreamRouteKind,
+    UpstreamStream,
+};
 pub mod middle_proxy;

src/transport/shadowsocks.rs

@@ -0,0 +1,60 @@
+use std::net::{IpAddr, SocketAddr};
+use std::time::Duration;
+
+use shadowsocks::{
+    ProxyClientStream,
+    config::{ServerConfig, ServerType},
+    context::Context,
+    net::ConnectOpts,
+};
+
+use crate::error::{ProxyError, Result};
+
+pub(crate) type ShadowsocksStream = ProxyClientStream<shadowsocks::net::TcpStream>;
+
+fn parse_server_config(url: &str, connect_timeout: Duration) -> Result<ServerConfig> {
+    let mut config = ServerConfig::from_url(url)
+        .map_err(|error| ProxyError::Config(format!("invalid shadowsocks url: {error}")))?;
+
+    if config.plugin().is_some() {
+        return Err(ProxyError::Config(
+            "shadowsocks plugins are not supported".to_string(),
+        ));
+    }
+
+    config.set_timeout(connect_timeout);
+    Ok(config)
+}
+
+pub(crate) fn sanitize_shadowsocks_url(url: &str) -> Result<String> {
+    Ok(parse_server_config(url, Duration::from_secs(1))?
+        .addr()
+        .to_string())
+}
+
+fn connect_opts_for_interface(interface: &Option<String>) -> ConnectOpts {
+    let mut opts = ConnectOpts::default();
+    if let Some(interface) = interface {
+        if let Ok(ip) = interface.parse::<IpAddr>() {
+            opts.bind_local_addr = Some(SocketAddr::new(ip, 0));
+        } else {
+            opts.bind_interface = Some(interface.clone());
+        }
+    }
+    opts
+}
+
+pub(crate) async fn connect_shadowsocks(
+    url: &str,
+    interface: &Option<String>,
+    target: SocketAddr,
+    connect_timeout: Duration,
+) -> Result<ShadowsocksStream> {
+    let config = parse_server_config(url, connect_timeout)?;
+    let context = Context::new_shared(ServerType::Local);
+    let opts = connect_opts_for_interface(interface);
+
+    ProxyClientStream::connect_with_opts(context, &config, target, &opts)
+        .await
+        .map_err(ProxyError::Io)
+}

src/transport/upstream.rs

@@ -4,22 +4,28 @@
 
 #![allow(deprecated)]
 
+use rand::Rng;
 use std::collections::{BTreeSet, HashMap};
-use std::net::{SocketAddr, IpAddr};
+use std::net::{IpAddr, SocketAddr};
+use std::pin::Pin;
 use std::sync::Arc;
 use std::sync::atomic::{AtomicU64, AtomicUsize, Ordering};
+use std::task::{Context, Poll};
 use std::time::Duration;
+use tokio::io::{AsyncRead, AsyncWrite, ReadBuf};
 use tokio::net::TcpStream;
 use tokio::sync::RwLock;
 use tokio::time::Instant;
-use rand::Rng;
+use tracing::{debug, info, trace, warn};
-use tracing::{debug, warn, info, trace};
 
 use crate::config::{UpstreamConfig, UpstreamType};
-use crate::error::{Result, ProxyError};
+use crate::error::{ProxyError, Result};
 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::protocol::constants::{TG_DATACENTER_PORT, TG_DATACENTERS_V4, TG_DATACENTERS_V6};
 use crate::stats::Stats;
+use crate::transport::shadowsocks::{
+    ShadowsocksStream, connect_shadowsocks, sanitize_shadowsocks_url,
+};
 use crate::transport::socket::{create_outgoing_socket_bound, resolve_interface_ip};
 use crate::transport::socks::{connect_socks4, connect_socks5};
 
@@ -47,7 +53,10 @@ struct LatencyEma {
 
 impl LatencyEma {
     const fn new(alpha: f64) -> Self {
-        Self { value_ms: None, alpha }
+        Self {
+            value_ms: None,
+            alpha,
+        }
     }
 
     fn update(&mut self, sample_ms: f64) {
@@ -131,11 +140,17 @@ impl UpstreamState {
             return Some(ms);
         }
 
-        let (sum, count) = self.dc_latency.iter()
+        let (sum, count) = self
+            .dc_latency
+            .iter()
             .filter_map(|l| l.get())
             .fold((0.0, 0u32), |(s, c), v| (s + v, c + 1));
 
-        if count > 0 { Some(sum / count as f64) } else { None }
+        if count > 0 {
+            Some(sum / count as f64)
+        } else {
+            None
+        }
     }
 }
 
@@ -158,11 +173,78 @@ pub struct StartupPingResult {
     pub both_available: bool,
 }
 
+pub enum UpstreamStream {
+    Tcp(TcpStream),
+    Shadowsocks(Box<ShadowsocksStream>),
+}
+
+impl std::fmt::Debug for UpstreamStream {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        match self {
+            Self::Tcp(_) => f.write_str("UpstreamStream::Tcp(..)"),
+            Self::Shadowsocks(_) => f.write_str("UpstreamStream::Shadowsocks(..)"),
+        }
+    }
+}
+
+impl UpstreamStream {
+    pub fn into_tcp(self) -> Result<TcpStream> {
+        match self {
+            Self::Tcp(stream) => Ok(stream),
+            Self::Shadowsocks(_) => Err(ProxyError::Config(
+                "shadowsocks upstreams are not supported when general.use_middle_proxy = true"
+                    .to_string(),
+            )),
+        }
+    }
+}
+
+impl AsyncRead for UpstreamStream {
+    fn poll_read(
+        self: Pin<&mut Self>,
+        cx: &mut Context<'_>,
+        buf: &mut ReadBuf<'_>,
+    ) -> Poll<std::io::Result<()>> {
+        match self.get_mut() {
+            Self::Tcp(stream) => Pin::new(stream).poll_read(cx, buf),
+            Self::Shadowsocks(stream) => Pin::new(stream.as_mut()).poll_read(cx, buf),
+        }
+    }
+}
+
+impl AsyncWrite for UpstreamStream {
+    fn poll_write(
+        self: Pin<&mut Self>,
+        cx: &mut Context<'_>,
+        buf: &[u8],
+    ) -> Poll<std::io::Result<usize>> {
+        match self.get_mut() {
+            Self::Tcp(stream) => Pin::new(stream).poll_write(cx, buf),
+            Self::Shadowsocks(stream) => Pin::new(stream.as_mut()).poll_write(cx, buf),
+        }
+    }
+
+    fn poll_flush(self: Pin<&mut Self>, cx: &mut Context<'_>) -> Poll<std::io::Result<()>> {
+        match self.get_mut() {
+            Self::Tcp(stream) => Pin::new(stream).poll_flush(cx),
+            Self::Shadowsocks(stream) => Pin::new(stream.as_mut()).poll_flush(cx),
+        }
+    }
+
+    fn poll_shutdown(self: Pin<&mut Self>, cx: &mut Context<'_>) -> Poll<std::io::Result<()>> {
+        match self.get_mut() {
+            Self::Tcp(stream) => Pin::new(stream).poll_shutdown(cx),
+            Self::Shadowsocks(stream) => Pin::new(stream.as_mut()).poll_shutdown(cx),
+        }
+    }
+}
+
 #[derive(Debug, Clone, Copy, PartialEq, Eq)]
 pub enum UpstreamRouteKind {
     Direct,
     Socks4,
     Socks5,
+    Shadowsocks,
 }
 
 #[derive(Debug, Clone)]
@@ -194,6 +276,7 @@ pub struct UpstreamApiSummarySnapshot {
     pub direct_total: usize,
     pub socks4_total: usize,
     pub socks5_total: usize,
+    pub shadowsocks_total: usize,
 }
 
 #[derive(Debug, Clone)]
@@ -253,7 +336,8 @@ impl UpstreamManager {
         connect_failfast_hard_errors: bool,
         stats: Arc<Stats>,
     ) -> Self {
-        let states = configs.into_iter()
+        let states = configs
+            .into_iter()
             .filter(|c| c.enabled)
             .map(UpstreamState::new)
             .collect();
@@ -311,20 +395,13 @@ impl UpstreamManager {
                 summary.unhealthy_total += 1;
             }
 
-            let (route_kind, address) = match &upstream.config.upstream_type {
+            let (route_kind, address) = Self::describe_upstream(&upstream.config.upstream_type);
-                UpstreamType::Direct { .. } => {
+            match route_kind {
-                    summary.direct_total += 1;
+                UpstreamRouteKind::Direct => summary.direct_total += 1,
-                    (UpstreamRouteKind::Direct, "direct".to_string())
+                UpstreamRouteKind::Socks4 => summary.socks4_total += 1,
-                }
+                UpstreamRouteKind::Socks5 => summary.socks5_total += 1,
-                UpstreamType::Socks4 { address, .. } => {
+                UpstreamRouteKind::Shadowsocks => summary.shadowsocks_total += 1,
-                    summary.socks4_total += 1;
+            }
-                    (UpstreamRouteKind::Socks4, address.clone())
-                }
-                UpstreamType::Socks5 { address, .. } => {
-                    summary.socks5_total += 1;
-                    (UpstreamRouteKind::Socks5, address.clone())
-                }
-            };
 
             let mut dc = Vec::with_capacity(NUM_DCS);
             for dc_idx in 0..NUM_DCS {
@@ -352,6 +429,18 @@ impl UpstreamManager {
         Some(UpstreamApiSnapshot { summary, upstreams })
     }
 
+    fn describe_upstream(upstream_type: &UpstreamType) -> (UpstreamRouteKind, String) {
+        match upstream_type {
+            UpstreamType::Direct { .. } => (UpstreamRouteKind::Direct, "direct".to_string()),
+            UpstreamType::Socks4 { address, .. } => (UpstreamRouteKind::Socks4, address.clone()),
+            UpstreamType::Socks5 { address, .. } => (UpstreamRouteKind::Socks5, address.clone()),
+            UpstreamType::Shadowsocks { url, .. } => (
+                UpstreamRouteKind::Shadowsocks,
+                sanitize_shadowsocks_url(url).unwrap_or_else(|_| "invalid".to_string()),
+            ),
+        }
+    }
+
     pub fn api_policy_snapshot(&self) -> UpstreamApiPolicySnapshot {
         UpstreamApiPolicySnapshot {
             connect_retry_attempts: self.connect_retry_attempts,
@@ -539,44 +628,44 @@ impl UpstreamManager {
         // Scope filter:
         //   If scope is set: only scoped and matched items
         //   If scope is not set: only unscoped items
-        let filtered_upstreams : Vec<usize> = upstreams.iter()
+        let filtered_upstreams: Vec<usize> = upstreams
+            .iter()
             .enumerate()
             .filter(|(_, u)| {
-                scope.map_or(
+                scope.map_or(u.config.scopes.is_empty(), |req_scope| {
-                    u.config.scopes.is_empty(),
+                    u.config
-                    |req_scope| {
+                        .scopes
-                        u.config.scopes
+                        .split(',')
-                            .split(',')
+                        .map(str::trim)
-                            .map(str::trim)
+                        .any(|s| s == req_scope)
-                            .any(|s| s == req_scope)
+                })
-                    }
-                )
             })
             .map(|(i, _)| i)
             .collect();
 
         // Healthy filter
-        let healthy: Vec<usize> = filtered_upstreams.iter()
+        let healthy: Vec<usize> = filtered_upstreams
+            .iter()
             .filter(|&&i| upstreams[i].healthy)
             .copied()
             .collect();
 
         if filtered_upstreams.is_empty() {
-            if Self::should_emit_warn(
+            if Self::should_emit_warn(self.no_upstreams_warn_epoch_ms.as_ref(), 5_000) {
-                self.no_upstreams_warn_epoch_ms.as_ref(),
+                warn!(
-                5_000,
+                    scope = scope,
-            ) {
+                    "No upstreams available! Using first (direct?)"
-                warn!(scope = scope, "No upstreams available! Using first (direct?)");
+                );
             }
             return None;
         }
 
         if healthy.is_empty() {
-            if Self::should_emit_warn(
+            if Self::should_emit_warn(self.no_healthy_warn_epoch_ms.as_ref(), 5_000) {
-                self.no_healthy_warn_epoch_ms.as_ref(),
+                warn!(
-                5_000,
+                    scope = scope,
-            ) {
+                    "No healthy upstreams available! Using random."
-                warn!(scope = scope, "No healthy upstreams available! Using random.");
+                );
             }
             return Some(filtered_upstreams[rand::rng().gen_range(0..filtered_upstreams.len())]);
         }
@@ -585,14 +674,18 @@ impl UpstreamManager {
             return Some(healthy[0]);
         }
 
-        let weights: Vec<(usize, f64)> = healthy.iter().map(|&i| {
+        let weights: Vec<(usize, f64)> = healthy
-            let base = upstreams[i].config.weight as f64;
+            .iter()
-            let latency_factor = upstreams[i].effective_latency(dc_idx)
+            .map(|&i| {
-                .map(|ms| if ms > 1.0 { 1000.0 / ms } else { 1000.0 })
+                let base = upstreams[i].config.weight as f64;
-                .unwrap_or(1.0);
+                let latency_factor = upstreams[i]
+                    .effective_latency(dc_idx)
+                    .map(|ms| if ms > 1.0 { 1000.0 / ms } else { 1000.0 })
+                    .unwrap_or(1.0);
 
-            (i, base * latency_factor)
+                (i, base * latency_factor)
-        }).collect();
+            })
+            .collect();
 
         let total: f64 = weights.iter().map(|(_, w)| w).sum();
 
@@ -620,8 +713,34 @@ 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> {
+    pub async fn connect(
-        let (stream, _) = self.connect_with_details(target, dc_idx, scope).await?;
+        &self,
+        target: SocketAddr,
+        dc_idx: Option<i16>,
+        scope: Option<&str>,
+    ) -> Result<UpstreamStream> {
+        let idx = self
+            .select_upstream(dc_idx, scope)
+            .await
+            .ok_or_else(|| ProxyError::Config("No upstreams available".to_string()))?;
+
+        let mut upstream = {
+            let guard = self.upstreams.read().await;
+            guard[idx].config.clone()
+        };
+
+        if let Some(s) = scope {
+            upstream.selected_scope = s.to_string();
+        }
+
+        let bind_rr = {
+            let guard = self.upstreams.read().await;
+            guard.get(idx).map(|u| u.bind_rr.clone())
+        };
+
+        let (stream, _) = self
+            .connect_selected_upstream(idx, upstream, target, dc_idx, bind_rr)
+            .await?;
         Ok(stream)
     }
 
@@ -632,7 +751,9 @@ impl UpstreamManager {
         dc_idx: Option<i16>,
         scope: Option<&str>,
     ) -> Result<(TcpStream, UpstreamEgressInfo)> {
-        let idx = self.select_upstream(dc_idx, scope).await
+        let idx = self
+            .select_upstream(dc_idx, scope)
+            .await
             .ok_or_else(|| ProxyError::Config("No upstreams available".to_string()))?;
 
         let mut upstream = {
@@ -650,6 +771,20 @@ impl UpstreamManager {
             guard.get(idx).map(|u| u.bind_rr.clone())
         };
 
+        let (stream, egress) = self
+            .connect_selected_upstream(idx, upstream, target, dc_idx, bind_rr)
+            .await?;
+        Ok((stream.into_tcp()?, egress))
+    }
+
+    async fn connect_selected_upstream(
+        &self,
+        idx: usize,
+        upstream: UpstreamConfig,
+        target: SocketAddr,
+        dc_idx: Option<i16>,
+        bind_rr: Option<Arc<AtomicUsize>>,
+    ) -> Result<(UpstreamStream, UpstreamEgressInfo)> {
         let connect_started_at = Instant::now();
         let mut last_error: Option<ProxyError> = None;
         let mut attempts_used = 0u32;
@@ -662,8 +797,8 @@ impl UpstreamManager {
                 break;
             }
             let remaining_budget = self.connect_budget.saturating_sub(elapsed);
-            let attempt_timeout = Duration::from_secs(DIRECT_CONNECT_TIMEOUT_SECS)
+            let attempt_timeout =
-                .min(remaining_budget);
+                Duration::from_secs(DIRECT_CONNECT_TIMEOUT_SECS).min(remaining_budget);
             if attempt_timeout.is_zero() {
                 last_error = Some(ProxyError::ConnectionTimeout {
                     addr: target.to_string(),
@@ -786,9 +921,12 @@ impl UpstreamManager {
         target: SocketAddr,
         bind_rr: Option<Arc<AtomicUsize>>,
         connect_timeout: Duration,
-    ) -> Result<(TcpStream, UpstreamEgressInfo)> {
+    ) -> Result<(UpstreamStream, UpstreamEgressInfo)> {
         match &config.upstream_type {
-            UpstreamType::Direct { interface, bind_addresses } => {
+            UpstreamType::Direct {
+                interface,
+                bind_addresses,
+            } => {
                 let bind_ip = Self::resolve_bind_address(
                     interface,
                     bind_addresses,
@@ -796,9 +934,7 @@ impl UpstreamManager {
                     bind_rr.as_deref(),
                     true,
                 );
-                if bind_ip.is_none()
+                if bind_ip.is_none() && bind_addresses.as_ref().is_some_and(|v| !v.is_empty()) {
-                    && bind_addresses.as_ref().is_some_and(|v| !v.is_empty())
-                {
                     return Err(ProxyError::Config(format!(
                         "No valid bind_addresses for target family {target}"
                     )));
@@ -813,8 +949,10 @@ impl UpstreamManager {
 
                 socket.set_nonblocking(true)?;
                 match socket.connect(&target.into()) {
-                    Ok(()) => {},
+                    Ok(()) => {}
-                    Err(err) if err.raw_os_error() == Some(libc::EINPROGRESS) || err.kind() == std::io::ErrorKind::WouldBlock => {},
+                    Err(err)
+                        if err.raw_os_error() == Some(libc::EINPROGRESS)
+                            || err.kind() == std::io::ErrorKind::WouldBlock => {}
                     Err(err) => return Err(ProxyError::Io(err)),
                 }
 
@@ -836,7 +974,7 @@ impl UpstreamManager {
 
                 let local_addr = stream.local_addr().ok();
                 Ok((
-                    stream,
+                    UpstreamStream::Tcp(stream),
                     UpstreamEgressInfo {
                         upstream_id,
                         route_kind: UpstreamRouteKind::Direct,
@@ -846,8 +984,12 @@ impl UpstreamManager {
                         socks_proxy_addr: None,
                     },
                 ))
-            },
+            }
-            UpstreamType::Socks4 { address, interface, user_id } => {
+            UpstreamType::Socks4 {
+                address,
+                interface,
+                user_id,
+            } => {
                 // Try to parse as SocketAddr first (IP:port), otherwise treat as hostname:port
                 let mut stream = if let Ok(proxy_addr) = address.parse::<SocketAddr>() {
                     // IP:port format - use socket with optional interface binding
@@ -863,8 +1005,10 @@ impl UpstreamManager {
 
                     socket.set_nonblocking(true)?;
                     match socket.connect(&proxy_addr.into()) {
-                        Ok(()) => {},
+                        Ok(()) => {}
-                        Err(err) if err.raw_os_error() == Some(libc::EINPROGRESS) || err.kind() == std::io::ErrorKind::WouldBlock => {},
+                        Err(err)
+                            if err.raw_os_error() == Some(libc::EINPROGRESS)
+                                || err.kind() == std::io::ErrorKind::WouldBlock => {}
                         Err(err) => return Err(ProxyError::Io(err)),
                     }
 
@@ -888,14 +1032,16 @@ impl UpstreamManager {
                     // Hostname:port format - use tokio DNS resolution
                     // Note: interface binding is not supported for hostnames
                     if interface.is_some() {
-                        warn!("SOCKS4 interface binding is not supported for hostname addresses, ignoring");
+                        warn!(
+                            "SOCKS4 interface binding is not supported for hostname addresses, ignoring"
+                        );
                     }
                     Self::connect_hostname_with_dns_override(address, connect_timeout).await?
                 };
 
                 // replace socks user_id with config.selected_scope, if set
-                let scope: Option<&str> = Some(config.selected_scope.as_str())
+                let scope: Option<&str> =
-                    .filter(|s| !s.is_empty());
+                    Some(config.selected_scope.as_str()).filter(|s| !s.is_empty());
                 let _user_id: Option<&str> = scope.or(user_id.as_deref());
 
                 let bound = match tokio::time::timeout(
@@ -915,7 +1061,7 @@ impl UpstreamManager {
                 let local_addr = stream.local_addr().ok();
                 let socks_proxy_addr = stream.peer_addr().ok();
                 Ok((
-                    stream,
+                    UpstreamStream::Tcp(stream),
                     UpstreamEgressInfo {
                         upstream_id,
                         route_kind: UpstreamRouteKind::Socks4,
@@ -925,8 +1071,13 @@ impl UpstreamManager {
                         socks_proxy_addr,
                     },
                 ))
-            },
+            }
-            UpstreamType::Socks5 { address, interface, username, password } => {
+            UpstreamType::Socks5 {
+                address,
+                interface,
+                username,
+                password,
+            } => {
                 // Try to parse as SocketAddr first (IP:port), otherwise treat as hostname:port
                 let mut stream = if let Ok(proxy_addr) = address.parse::<SocketAddr>() {
                     // IP:port format - use socket with optional interface binding
@@ -942,8 +1093,10 @@ impl UpstreamManager {
 
                     socket.set_nonblocking(true)?;
                     match socket.connect(&proxy_addr.into()) {
-                        Ok(()) => {},
+                        Ok(()) => {}
-                        Err(err) if err.raw_os_error() == Some(libc::EINPROGRESS) || err.kind() == std::io::ErrorKind::WouldBlock => {},
+                        Err(err)
+                            if err.raw_os_error() == Some(libc::EINPROGRESS)
+                                || err.kind() == std::io::ErrorKind::WouldBlock => {}
                         Err(err) => return Err(ProxyError::Io(err)),
                     }
 
@@ -967,15 +1120,17 @@ impl UpstreamManager {
                     // Hostname:port format - use tokio DNS resolution
                     // Note: interface binding is not supported for hostnames
                     if interface.is_some() {
-                        warn!("SOCKS5 interface binding is not supported for hostname addresses, ignoring");
+                        warn!(
+                            "SOCKS5 interface binding is not supported for hostname addresses, ignoring"
+                        );
                     }
                     Self::connect_hostname_with_dns_override(address, connect_timeout).await?
                 };
 
                 debug!(config = ?config, "Socks5 connection");
                 // replace socks user:pass with config.selected_scope, if set
-                let scope: Option<&str> = Some(config.selected_scope.as_str())
+                let scope: Option<&str> =
-                    .filter(|s| !s.is_empty());
+                    Some(config.selected_scope.as_str()).filter(|s| !s.is_empty());
                 let _username: Option<&str> = scope.or(username.as_deref());
                 let _password: Option<&str> = scope.or(password.as_deref());
 
@@ -996,7 +1151,7 @@ impl UpstreamManager {
                 let local_addr = stream.local_addr().ok();
                 let socks_proxy_addr = stream.peer_addr().ok();
                 Ok((
-                    stream,
+                    UpstreamStream::Tcp(stream),
                     UpstreamEgressInfo {
                         upstream_id,
                         route_kind: UpstreamRouteKind::Socks5,
@@ -1006,7 +1161,22 @@ impl UpstreamManager {
                         socks_proxy_addr,
                     },
                 ))
-            },
+            }
+            UpstreamType::Shadowsocks { url, interface } => {
+                let stream = connect_shadowsocks(url, interface, target, connect_timeout).await?;
+                let local_addr = stream.get_ref().local_addr().ok();
+                    Ok((
+                        UpstreamStream::Shadowsocks(Box::new(stream)),
+                        UpstreamEgressInfo {
+                        upstream_id,
+                        route_kind: UpstreamRouteKind::Shadowsocks,
+                        local_addr,
+                        direct_bind_ip: None,
+                        socks_bound_addr: None,
+                        socks_proxy_addr: None,
+                    },
+                ))
+            }
         }
     }
 
@@ -1023,7 +1193,9 @@ impl UpstreamManager {
     ) -> Vec<StartupPingResult> {
         let upstreams: Vec<(usize, UpstreamConfig, Arc<AtomicUsize>)> = {
             let guard = self.upstreams.read().await;
-            guard.iter().enumerate()
+            guard
+                .iter()
+                .enumerate()
                 .map(|(i, u)| (i, u.config.clone(), u.bind_rr.clone()))
                 .collect()
         };
@@ -1051,6 +1223,11 @@ impl UpstreamManager {
                 }
                 UpstreamType::Socks4 { address, .. } => format!("socks4://{}", address),
                 UpstreamType::Socks5 { address, .. } => format!("socks5://{}", address),
+                UpstreamType::Shadowsocks { url, .. } => {
+                    let address =
+                        sanitize_shadowsocks_url(url).unwrap_or_else(|_| "invalid".to_string());
+                    format!("shadowsocks://{address}")
+                }
             };
 
             let mut v6_results = Vec::with_capacity(NUM_DCS);
@@ -1061,8 +1238,14 @@ impl UpstreamManager {
 
                     let result = tokio::time::timeout(
                         Duration::from_secs(DC_PING_TIMEOUT_SECS),
-                        self.ping_single_dc(*upstream_idx, upstream_config, Some(bind_rr.clone()), addr_v6)
+                        self.ping_single_dc(
-                    ).await;
+                            *upstream_idx,
+                            upstream_config,
+                            Some(bind_rr.clone()),
+                            addr_v6,
+                        ),
+                    )
+                    .await;
 
                     let ping_result = match result {
                         Ok(Ok(rtt_ms)) => {
@@ -1112,8 +1295,14 @@ impl UpstreamManager {
 
                     let result = tokio::time::timeout(
                         Duration::from_secs(DC_PING_TIMEOUT_SECS),
-                        self.ping_single_dc(*upstream_idx, upstream_config, Some(bind_rr.clone()), addr_v4)
+                        self.ping_single_dc(
-                    ).await;
+                            *upstream_idx,
+                            upstream_config,
+                            Some(bind_rr.clone()),
+                            addr_v4,
+                        ),
+                    )
+                    .await;
 
                     let ping_result = match result {
                         Ok(Ok(rtt_ms)) => {
@@ -1162,7 +1351,7 @@ impl UpstreamManager {
                     Err(_) => {
                         warn!(dc = %dc_key, "Invalid dc_overrides key, skipping");
                         continue;
-                    },
+                    }
                     _ => continue,
                 };
                 let dc_idx = dc_num as usize;
@@ -1175,8 +1364,14 @@ impl UpstreamManager {
                             }
                             let result = tokio::time::timeout(
                                 Duration::from_secs(DC_PING_TIMEOUT_SECS),
-                                self.ping_single_dc(*upstream_idx, upstream_config, Some(bind_rr.clone()), addr)
+                                self.ping_single_dc(
-                            ).await;
+                                    *upstream_idx,
+                                    upstream_config,
+                                    Some(bind_rr.clone()),
+                                    addr,
+                                ),
+                            )
+                            .await;
 
                             let ping_result = match result {
                                 Ok(Ok(rtt_ms)) => DcPingResult {
@@ -1205,7 +1400,9 @@ impl UpstreamManager {
                                 v4_results.push(ping_result);
                             }
                         }
-                        Err(_) => warn!(dc = %dc_idx, addr = %addr_str, "Invalid dc_overrides address, skipping"),
+                        Err(_) => {
+                            warn!(dc = %dc_idx, addr = %addr_str, "Invalid dc_overrides address, skipping")
+                        }
                     }
                 }
             }
@@ -1381,12 +1578,8 @@ impl UpstreamManager {
         ipv6_enabled: bool,
         dc_overrides: HashMap<String, Vec<String>>,
     ) {
-        let groups = Self::build_health_check_groups(
+        let groups =
-            prefer_ipv6,
+            Self::build_health_check_groups(prefer_ipv6, ipv4_enabled, ipv6_enabled, &dc_overrides);
-            ipv4_enabled,
-            ipv6_enabled,
-            &dc_overrides,
-        );
         let required_healthy_groups = Self::required_healthy_group_count(groups.len());
         let mut endpoint_rotation: HashMap<(usize, i16, bool), usize> = HashMap::new();
 
@@ -1416,13 +1609,16 @@ impl UpstreamManager {
                     let mut group_ok = false;
                     let mut group_rtt_ms = None;
 
-                    for (is_primary, endpoints) in [(true, &group.primary), (false, &group.fallback)] {
+                    for (is_primary, endpoints) in
+                        [(true, &group.primary), (false, &group.fallback)]
+                    {
                         if endpoints.is_empty() {
                             continue;
                         }
 
                         let rotation_key = (i, group.dc_idx, is_primary);
-                        let start_idx = *endpoint_rotation.entry(rotation_key).or_insert(0) % endpoints.len();
+                        let start_idx =
+                            *endpoint_rotation.entry(rotation_key).or_insert(0) % endpoints.len();
                         let mut next_idx = (start_idx + 1) % endpoints.len();
 
                         for step in 0..endpoints.len() {
@@ -1544,8 +1740,7 @@ impl UpstreamManager {
             return None;
         }
 
-        UpstreamState::dc_array_idx(dc_idx)
+        UpstreamState::dc_array_idx(dc_idx).map(|idx| guard[0].dc_ip_pref[idx])
-            .map(|idx| guard[0].dc_ip_pref[idx])
     }
 
     /// Get preferred DC address based on config preference
@@ -1566,6 +1761,12 @@ impl UpstreamManager {
 #[cfg(test)]
 mod tests {
     use super::*;
+    use std::sync::Arc;
+
+    use crate::stats::Stats;
+
+    const TEST_SHADOWSOCKS_URL: &str =
+        "ss://2022-blake3-aes-256-gcm:MDEyMzQ1Njc4OTAxMjM0NTY3ODkwMTIzNDU2Nzg5MDE=@127.0.0.1:8388";
 
     #[test]
     fn required_healthy_group_count_applies_three_group_threshold() {
@@ -1596,15 +1797,18 @@ mod tests {
 
         assert!(dc2.primary.iter().all(|addr| addr.is_ipv6()));
         assert!(dc2.fallback.iter().all(|addr| addr.is_ipv4()));
-        assert!(dc2
+        assert!(
-            .primary
+            dc2.primary
-            .contains(&"[2001:db8::10]:443".parse::<SocketAddr>().unwrap()));
+                .contains(&"[2001:db8::10]:443".parse::<SocketAddr>().unwrap())
-        assert!(dc2
+        );
-            .fallback
+        assert!(
-            .contains(&"203.0.113.10:443".parse::<SocketAddr>().unwrap()));
+            dc2.fallback
-        assert!(dc2
+                .contains(&"203.0.113.10:443".parse::<SocketAddr>().unwrap())
-            .fallback
+        );
-            .contains(&"203.0.113.11:443".parse::<SocketAddr>().unwrap()));
+        assert!(
+            dc2.fallback
+                .contains(&"203.0.113.11:443".parse::<SocketAddr>().unwrap())
+        );
     }
 
     #[test]
@@ -1626,12 +1830,14 @@ mod tests {
             .expect("override-only dc group must be present");
 
         assert_eq!(dc9.primary.len(), 2);
-        assert!(dc9
+        assert!(
-            .primary
+            dc9.primary
-            .contains(&"198.51.100.1:443".parse::<SocketAddr>().unwrap()));
+                .contains(&"198.51.100.1:443".parse::<SocketAddr>().unwrap())
-        assert!(dc9
+        );
-            .primary
+        assert!(
-            .contains(&"198.51.100.2:443".parse::<SocketAddr>().unwrap()));
+            dc9.primary
+                .contains(&"198.51.100.2:443".parse::<SocketAddr>().unwrap())
+        );
         assert!(dc9.fallback.is_empty());
     }
 
@@ -1678,4 +1884,36 @@ mod tests {
 
         assert_eq!(bind, None);
     }
+
+    #[test]
+    fn api_snapshot_reports_shadowsocks_as_sanitized_route() {
+        let manager = UpstreamManager::new(
+            vec![UpstreamConfig {
+                upstream_type: UpstreamType::Shadowsocks {
+                    url: TEST_SHADOWSOCKS_URL.to_string(),
+                    interface: None,
+                },
+                weight: 2,
+                enabled: true,
+                scopes: String::new(),
+                selected_scope: String::new(),
+            }],
+            1,
+            100,
+            1000,
+            1,
+            false,
+            Arc::new(Stats::new()),
+        );
+
+        let snapshot = manager.try_api_snapshot().expect("snapshot");
+        assert_eq!(snapshot.summary.configured_total, 1);
+        assert_eq!(snapshot.summary.shadowsocks_total, 1);
+        assert_eq!(snapshot.upstreams.len(), 1);
+        assert_eq!(
+            snapshot.upstreams[0].route_kind,
+            UpstreamRouteKind::Shadowsocks
+        );
+        assert_eq!(snapshot.upstreams[0].address, "127.0.0.1:8388");
+    }
 }