astelm/minio
1
0
Fork 0
mirror of https://github.com/pgsty/minio.git synced 2026-11-09 17:21:51 +03:00
Code Issues Packages Projects Releases Wiki Activity

fix(restore): preserve encoded object identity and part layout

Browse Source 
Signed-off-by: Feng Ruohang <rh@vonng.com>
This commit is contained in:
Feng Ruohang
2026-11-03 22:16:47 +08:00
parent 50a234d6c3
commit 4f72d0ea94
2 changed files with 211 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files
cmd/erasure-object.go
+16 -4
View File
@@ -2687,14 +2687,23 @@ func (er erasureObjects) restoreTransitionedObject(ctx context.Context, bucket s
oi = actualfi.ToObjectInfo(bucket, object, opts.Versioned || opts.VersionSuspended)
ropts := putRestoreOpts(bucket, object, opts.Transition.RestoreRequest, oi)
ropts.NoLock = opts.NoLock
if len(oi.Parts) == 1 {
// Rehydration copies encoded bytes without changing their content identity.
ropts.PreserveETag = oi.ETag
// Even a one-part multipart upload must retain its original part number:
// encrypted multipart data uses that number to derive its decryption key.
if len(oi.Parts) == 1 && !oi.isMultipart() {
actualSize, err := oi.GetActualSize()
if err != nil {
return setRestoreHeaderFn(oi, err)
}
ropts.IndexCB = func() []byte { return oi.Parts[0].Index }
var rs *HTTPRangeSpec
gr, err := getTransitionedObjectReader(ctx, bucket, object, rs, http.Header{}, oi, opts)
if err != nil {
return setRestoreHeaderFn(oi, toObjectErr(err, bucket, object))
}
defer gr.Close()
hashReader, err := hash.NewReader(ctx, gr, gr.ObjInfo.Size, "", "", gr.ObjInfo.Size)
hashReader, err := hash.NewReader(ctx, gr, gr.ObjInfo.Size, "", "", actualSize)
if err != nil {
return setRestoreHeaderFn(oi, toObjectErr(err, bucket, object))
}
@@ -2719,11 +2728,14 @@ func (er erasureObjects) restoreTransitionedObject(ctx context.Context, bucket s
// rehydrate the parts back on disk as per the original xl.meta prior to transition
for _, partInfo := range oi.Parts {
hr, err := hash.NewReader(ctx, io.LimitReader(gr, partInfo.Size), partInfo.Size, "", "", partInfo.Size)
// Stored and logical lengths differ for compressed/encrypted data.
hr, err := hash.NewReader(ctx, io.LimitReader(gr, partInfo.Size), partInfo.Size, "", "", partInfo.ActualSize)
if err != nil {
return setRestoreHeaderFn(oi, err)
}
pInfo, err := er.PutObjectPart(ctx, bucket, object, res.UploadID, partInfo.Number, NewPutObjReader(hr), ObjectOptions{})
pInfo, err := er.PutObjectPart(ctx, bucket, object, res.UploadID, partInfo.Number, NewPutObjReader(hr), ObjectOptions{
IndexCB: func() []byte { return partInfo.Index },
})
if err != nil {
return setRestoreHeaderFn(oi, err)
}
cmd/restore-native-encoding_test.go
+195
View File
@@ -0,0 +1,195 @@
// Copyright (c) 2026 Feng Ruohang
//
// This file is part of Silo Object Storage stack
//
// This program is free software: you can redistribute it and/or modify
// it under the terms of the GNU Affero General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.
//
// This program is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU Affero General Public License for more details.
//
// You should have received a copy of the GNU Affero General Public License
// along with this program. If not, see <http://www.gnu.org/licenses/>.
package cmd
import (
"bytes"
"fmt"
"net/http"
"net/http/httptest"
"net/url"
"testing"
"time"
"github.com/minio/minio/internal/bucket/versioning"
"github.com/minio/minio/internal/event"
"github.com/minio/minio/internal/hash"
xhttp "github.com/minio/minio/internal/http"
"github.com/minio/minio/internal/kms"
"github.com/minio/minio/internal/pubsub"
"github.com/minio/mux"
)
// Exercise the stored representation produced by real S3 uploads. Native
// restore copies encoded bytes, but must retain logical sizes and identity.
func TestNativeRestoreEncodedObjects(t *testing.T) {
for _, pools := range []int{1, 2} {
for _, encoding := range []string{"plain", "compressed", "ssec", "s3-compressed"} {
for _, layout := range []string{"single", "multipart", "sparse-one-part"} {
t.Run(fmt.Sprintf("pools=%d/%s/%s", pools, encoding, layout), func(t *testing.T) {
r := newRestoreSafetyTest(t, pools, versioning.Enabled)
initConfigSubsystem(t.Context(), r.z)
r.object = "restore-encoded.txt"
previousKMS, previousTLS := GlobalKMS, globalIsTLS
t.Cleanup(func() { GlobalKMS, globalIsTLS = previousKMS, previousTLS })
globalIsTLS = true
headers := make(map[string]string)
readHeaders := make(map[string]string)
if encoding == "compressed" || encoding == "s3-compressed" {
enableAttributesPartsCompression(t)
}
if encoding == "ssec" {
headers = ssecTestHeaders(0x53)
readHeaders = ssecTestHeaders(0x53)
}
if encoding == "s3-compressed" {
var err error
GlobalKMS, err = kms.NewBuiltin("restore-test-key", bytes.Repeat([]byte{0x59}, 32))
if err != nil {
t.Fatal(err)
}
headers[xhttp.AmzServerSideEncryption] = xhttp.AmzEncryptionAES
}
previousObjectAPI := newObjectLayerFn()
t.Cleanup(func() { setObjectLayer(previousObjectAPI) })
router := initTestAPIEndPoints(r.z, []string{"NewMultipart", "PutObjectPart", "CompleteMultipart", "PutObject", "GetObject"})
bodies := [][]byte{bytes.Repeat([]byte("encoded-restore-"), compMinIndexSize/16+1)}
partNumbers := []int{1}
switch layout {
case "multipart":
bodies = append(bodies, []byte("last encoded part"))
partNumbers = []int{1, 2}
case "sparse-one-part":
partNumbers = []int{5}
}
want := bytes.Join(bodies, nil)
if layout == "single" {
headers[xhttp.AmzChecksumCRC32] = mustChecksum(t, hash.ChecksumCRC32, want)
putCopyChecksumSource(t, router, globalActiveCred, r.bucket, r.object, want, headers)
} else {
attributesPartsUpload(t, router, globalActiveCred, r.bucket, r.object, headers, bodies, partNumbers)
}
before, err := r.z.GetObjectInfo(t.Context(), r.bucket, r.object, ObjectOptions{})
if err != nil {
t.Fatal(err)
}
if before.IsCompressed() != (encoding == "compressed" || encoding == "s3-compressed") {
t.Fatalf("unexpected compression state: %v", before.IsCompressed())
}
if before.IsCompressed() && len(before.Parts[0].Index) == 0 {
t.Fatal("fixture must have a compression index")
}
if layout == "single" && len(before.Checksum) == 0 {
t.Fatal("fixture must have a stored checksum")
}
objectURL := getGetObjectURL("", r.bucket, r.object) + "?versionId=" + url.QueryEscape(before.VersionID)
beforeGet := attributesPartsSignedRequest(t, router, globalActiveCred, http.MethodGet, objectURL, nil, readHeaders)
if beforeGet.Code != http.StatusOK || !bytes.Equal(beforeGet.Body.Bytes(), want) {
t.Fatalf("fixture GET: %d bytes=%d want=%d", beforeGet.Code, beforeGet.Body.Len(), len(want))
}
beforeRaw := nativeRestoreSnapshot(t, r)
topts := r.opts(before.VersionID)
topts.MTime = before.ModTime
topts.Transition = TransitionOptions{Tier: "WARM", ETag: before.ETag}
if err := r.z.TransitionObject(t.Context(), r.bucket, r.object, topts); err != nil {
t.Fatal(err)
}
observed := nativeRestoreHTTPObserver{r.z, make(chan nativeRestoreHTTPResult, 1)}
api := objectAPIHandlers{ObjectAPI: func() ObjectLayer { return observed }}
events, done := make(chan event.Event, 1), make(chan struct{})
defer close(done)
if err := globalHTTPListen.Subscribe(pubsub.MaskFromMaskable(event.ObjectRestoreCompleted), events, done, nil); err != nil {
t.Fatal(err)
}
restoreXML := []byte(`<RestoreRequest xmlns="http://s3.amazonaws.com/doc/2006-03-01/"><Days>1</Days></RestoreRequest>`)
req, err := newTestSignedRequestV4(http.MethodPost, objectURL+"&restore", int64(len(restoreXML)), bytes.NewReader(restoreXML), globalActiveCred.AccessKey, globalActiveCred.SecretKey, nil)
if err != nil {
t.Fatal(err)
}
if err := req.ParseForm(); err != nil {
t.Fatal(err)
}
req = mux.SetURLVars(req, map[string]string{"bucket": r.bucket, "object": r.object})
rec := httptest.NewRecorder()
api.PostRestoreObjectHandler(rec, req)
if rec.Code != http.StatusOK {
t.Fatalf("restore POST: %d %s", rec.Code, rec.Body.String())
}
select {
case result := <-observed.done:
if result.err != nil {
t.Fatalf("background restore: %v", result.err)
}
case <-time.After(5 * time.Second):
t.Fatal("restore did not finish")
}
select {
case <-events:
case <-time.After(5 * time.Second):
t.Fatal("restore completion event missing")
}
after, err := r.z.GetObjectInfo(t.Context(), r.bucket, r.object, r.opts(before.VersionID))
if err != nil {
t.Fatal(err)
}
if after.ETag != before.ETag || after.Size != before.Size || !after.ModTime.Equal(before.ModTime) {
t.Error("restore changed content identity")
}
afterRaw := nativeRestoreSnapshot(t, r)
for pool := range beforeRaw {
if len(afterRaw[pool]) != len(beforeRaw[pool]) {
t.Fatalf("pool %d version count changed", pool)
}
for i, fi := range beforeRaw[pool] {
if !bytes.Equal(fi.Checksum, afterRaw[pool][i].Checksum) {
t.Error("restore lost stored checksum")
}
}
}
if len(after.Parts) != len(before.Parts) {
t.Fatal("restore changed part count")
}
for i, part := range after.Parts {
old := before.Parts[i]
if part.Number != old.Number || part.Size != old.Size || part.ActualSize != old.ActualSize || !bytes.Equal(part.Index, old.Index) {
t.Errorf("part %d layout or compression index changed", i)
}
}
tier := globalTierConfigMgr.drivercache["WARM"].(*restoreTestTier)
reads := tier.gets.Load()
got := attributesPartsSignedRequest(t, router, globalActiveCred, http.MethodGet, objectURL, nil, readHeaders)
if got.Code != http.StatusOK || !bytes.Equal(got.Body.Bytes(), want) {
t.Errorf("restored GET: %d bytes=%d want=%d", got.Code, got.Body.Len(), len(want))
}
if got.Header().Get(xhttp.ETag) != beforeGet.Header().Get(xhttp.ETag) {
t.Error("restore changed client ETag")
}
for i, number := range partNumbers {
part := attributesPartsSignedRequest(t, router, globalActiveCred, http.MethodGet, objectURL+fmt.Sprintf("&partNumber=%d", number), nil, readHeaders)
if part.Code != http.StatusPartialContent || !bytes.Equal(part.Body.Bytes(), bodies[i]) {
t.Errorf("restored part %d: %d bytes=%d want=%d", number, part.Code, part.Body.Len(), len(bodies[i]))
}
}
if tier.gets.Load() != reads {
t.Error("restored GET still used remote tier")
}
})
}
}
}
}