astelm/minio
1
0
Fork 0
mirror of https://github.com/pgsty/minio.git synced 2026-11-10 17:45:32 +03:00
Code Issues Packages Projects Releases Wiki Activity

fix(restore): stage native restores outside the object lock

Browse Source 
The unreleased native-restore fixes after RELEASE.2026-10-26T00-00-00Z
let the lock decide who owns a restore. Since fa42a0982 a single-pool
worker holds the object lock for the whole tier download, and since
7f38bf628 a busy lock means "a restore runs". 7f38bf628 also keeps the
restore's data in a marked multipart upload. Two regressions followed:

- While an old version restored, every GET, HEAD and write of the same
  key waited for the download. On a single pool this is new; with
  several pools the released code already held the lock that long.
- An ordinary upload whose completion failed after staging its metadata
  became InvalidUploadID for every caller, and could not be aborted for
  an hour. The marker that hid restore uploads could not tell them apart
  once a completion had rewritten the upload's xl.meta.

A restore now runs in two phases:

- The worker erasure-codes the tier copy into the tmp volume of the
  version's set, without the object lock, as heal does. A drive that
  fails any part, or misses a part when CheckParts runs before the
  commit, takes no part in the commit.
- A restore lock per version tells whether a restore runs. POST takes it
  before the object lock and hands it to the worker. It is a liveness
  hint: only the checks under the object lock decide what is written.
- The commit takes the object lock briefly, on one pool or several. It
  reads the version again and checks the attempt's ownership, the
  version identity, the tier copy, the part layout and the erasure
  layout. It then publishes with RenameData drive by drive, never
  through erasure renameData's undo, and needs the write quorum. The
  replaced data directory goes only when every published drive reports
  the same one.
- A publication that reached some drives but not the write quorum
  converges under the same lock hold, in bounded retries after it: back
  to the unrestored version on every drive that holds it, or to a
  committed restore when the copy turns out to be on a write quorum.
  Each pass first checks that every drive holds what the version held
  before the commit, the commit wrote, or a convergence wrote, and stops
  on anything else. It deletes no directory.

Multipart goes back to main: erasure-multipart.go and
bucket-object-lock.go are identical to 69fcf50cd. The commit lock, the
marker and the restore exceptions of multipart completion and of the
null-version Object Lock check are gone, and single-pool completion
again takes the upload lock first and the object lock right before it
commits. With several pools, abandoned-data cleanup now takes the pools
lock that writes take, as heal does, so it cannot remove a directory
that a commit moved in place and has not yet referenced. The restore
capability becomes protocol 2, and a queued worker checks it again.

Recovery after a coordinator crash during publication, and storage RPCs
that complete after their caller returned, stay excluded. Main has the
same exposure, and they are tracked separately.

Tests:
- New regression tests cover both problems. They fail on f85d4a5fa,
  and the multipart one also passes on main.
- Fault-injection tests cover a publication below the write quorum,
  including seven of twelve drives on 8+4; lost answers; convergence
  retries and stops; the old data directory; a stale inline marker; a
  recreated tmp volume; a published revision read across its expiry; a
  second POST; a target deleted during the download; a tier 404; and
  pools-locked abandoned cleanup.
- Thirteen tests of the upload marker from 7f38bf628 are dropped, and
  so is TestNativeRestoreConcurrentCompletion from fa42a0982, which
  completed the restore's upload from a client. Four tests are reframed
  for the new semantics, including the lock-domain restore test, which
  now gates the commit.
- Three self-recursive WriteMetadata test doubles are fixed, and the
  helpers of the dropped tests are removed.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Feng Ruohang <rh@vonng.com>
This commit is contained in:
Feng Ruohang
2026-11-06 11:30:29 +08:00
parent f85d4a5fa1
commit 45b1e4bc00
17 changed files with 1956 additions and 2026 deletions
Download Patch File Download Diff File Expand all files Collapse all files
cmd/erasure-server-pool.go
+32 -54
View File
@@ -2393,22 +2393,21 @@ func (z *erasureServerPools) CompleteMultipartUpload(ctx context.Context, bucket
}
}()
// Keep object -> upload lock ordering for single-pool completion too:
// native restore holds the object lock while publishing upload parts.
// Both locks are in the object's lock domain, the set-local one of
// single-pool writes.
if !opts.NoLock {
lk := z.NewObjectNSLock(bucket, encodeDirObject(object))
lkctx, err := lk.GetLock(ctx, globalOperationTimeout)
if err != nil {
return objInfo, err
}
ctx = lkctx.Context()
defer lk.Unlock(lkctx)
}
opts.NoLock = true
// Several pools take the object lock first, then the upload lock below.
// A single pool takes the upload lock here and the set takes the object
// lock, in the object's own erasure set, only right before it commits.
if !z.SinglePool() {
if !opts.NoLock {
lk := z.NewObjectNSLock(bucket, encodeDirObject(object))
lkctx, err := lk.GetLock(ctx, globalOperationTimeout)
if err != nil {
return objInfo, err
}
ctx = lkctx.Context()
defer lk.Unlock(lkctx)
}
opts.NoLock = true
// A conditional completion must be evaluated against the logical
// latest object across pools, under the object write lock held for
// this operation. The pool hosting the upload may only hold a stale
@@ -2460,8 +2459,9 @@ func (z *erasureServerPools) CompleteMultipartUpload(ctx context.Context, bucket
// Hold write locks to verify uploaded parts, also disallows any
// parallel PutObjectPart() requests.
// An empty upload ID aliases the object lock already held above.
// Let the storage layer return its normal missing bucket/upload error.
// An empty upload ID would name the object lock itself, which this
// completion holds or takes later: let the storage layer return its
// normal missing bucket/upload error instead.
if uploadID != "" {
uploadIDLock := z.uploadIDNSLock(bucket, object, uploadID)
wlkctx, err := uploadIDLock.GetLock(ctx, globalOperationTimeout)
@@ -3398,48 +3398,26 @@ func (z *erasureServerPools) TransitionObject(ctx context.Context, bucket, objec
return z.serverPools[idx].TransitionObject(ctx, bucket, object, opts)
}
// RestoreTransitionedObject - restore transitioned object content locally on this cluster.
func (z *erasureServerPools) RestoreTransitionedObject(ctx context.Context, bucket, object string, opts ObjectOptions) (err error) {
object = encodeDirObject(object)
defer func() {
// The restore stopped before it could clear POST's ongoing marker.
var reset restoreResetError
if errors.As(err, &reset) {
z.recoverRestoreStatus(bucket, object, opts)
err = reset.error
}
}()
if z.SinglePool() {
return z.serverPools[0].RestoreTransitionedObject(ctx, bucket, object, opts)
}
if !opts.NoLock {
// Acquire write lock before restoring transitioned object
lk := z.NewNSLock(bucket, object)
lkctx, err := getRestoreLock(ctx, lk)
if err != nil {
return restoreResetError{err}
}
ctx = lkctx.Context()
defer lk.Unlock(lkctx)
}
// Avoid restoring object from a pool being decommissioned.
opts.SkipDecommissioned = true
opts.NoLock = true
idx, err := z.restorePoolIdx(ctx, bucket, object, opts)
if err != nil {
return err
}
return z.serverPools[idx].RestoreTransitionedObject(ctx, bucket, object, opts)
}
func (z *erasureServerPools) CheckAbandonedParts(ctx context.Context, bucket, object string, opts madmin.HealOpts) error {
object = encodeDirObject(object)
if z.SinglePool() {
return z.serverPools[0].CheckAbandonedParts(ctx, bucket, object, opts)
}
// Several pools: remove abandoned data under the pools lock, which every
// write of the object takes, as healObjectInPool does. The lock of one
// pool's set would not exclude a write that has moved its data directory
// in place but not yet published the metadata that references it, such
// as a native restore's commit.
if opts.Remove && !opts.DryRun && !opts.NoLock {
lk := z.NewObjectNSLock(bucket, object)
lkctx, err := lk.GetLock(ctx, globalOperationTimeout)
if err != nil {
return err
}
ctx = lkctx.Context()
defer lk.Unlock(lkctx)
opts.NoLock = true
}
errs := make([]error, len(z.serverPools))
var wg sync.WaitGroup
for idx, pool := range z.serverPools {