s3: evaluate all conditional write headers under the object lock (#12211)

* filer_pb,filer: per-mutation preconditions on ObjectTransaction

A transaction's request-level condition evaluates one entry before any
mutation runs, so a commit guarding two paths — the object it publishes
and the upload directory it assembles — could not be expressed. Give each
ObjectMutation an optional condition (evaluated against condition_key,
else the mutation's target, under the same lock just before the mutation
applies) and report which mutation failed through failed_mutation on the
response, so callers can tell a mutation guard's rejection apart from the
request-level precondition's.

* s3: evaluate multipart completion conditions under the object lock

CompleteMultipartUpload checked conditional headers gateway-side before
committing, and the routed commit transaction guarded only the upload
directory — nothing made If-None-Match: * atomic with the publish, so
concurrent conditional completions of the same absent key could all
report success.

The request's reducible write condition (If-None-Match/If-Match, star or
single strong ETag) now rides the commit as the transaction's request
condition on the owner filer: on the object path for unversioned buckets,
falling back to the .versions directory for suspended and enabled
versioning. The upload-exists guard moves onto the PUT mutation, keeping
its check-before-publish ordering. Conditions the reducer cannot express
stay on the distributed write lock, where check and commit already
serialize.

* filer,s3: resolve the latest version first for versioned conditional writes

* filer: keep directory entries holding content out of the absent-slot rule

* s3: keep a gateway-side upload-exists check for filers without mutation conditions

* filer: honor null-version-is-latest resolving prefer-alt conditions

* drop scratch multipart-race repro programs

* s3: note the write lock does not cover routed writes on non-reducible conditions

* filer: mirror the gateway's remote-etag precedence in condition evaluation

storedEntryETag only knew the extended attribute and the chunk fallback,
so a synced remote entry compared chunk-derived ETags while the gateway
compares the remote ETag. Follow the same precedence: extended attribute,
remote ETag while the entry still holds the remote bytes, Md5, then the
chunk hash — and no ETag for an entry holding none.

* s3: reduce all conditional write headers into owner-lock clauses

Time-based headers, combined conditions, ETag lists, and weak ETags stayed
on the distributed write lock because buildWriteCondition could not express
them, leaving their gateway check racing owner-routed writes. Reduce every
parseable header form to WriteCondition clauses — multiple clauses compose
AND, ETag candidates carry the raw list for the filer's canonicalization,
and RFC 7232 precedence keeps If-Unmodified-Since hidden by If-Match and
If-Modified-Since hidden by If-None-Match — so the owner filer evaluates
them under the same object-path lock as the commit. Only unparseable
headers still fall back, where the lock path reports the same parse error.
This commit is contained in:
Chris Lu authored and GitHub committed 2026-10-29 11:49:48 -07:00
1 parent 2737315e9d
commit 2dbf9c9108
7 files changed
+162 -56

No files matched your search

+3 -4
View File
@@ -691,10 +691,9 @@ func (s3a *S3ApiServer) completeMultipartUpload(r *http.Request, input *s3.Compl
routeKey := s3a.objectRouteKey(*input.Bucket, *input.Key)
// The write condition rides the commit transaction so the owner filer
// evaluates it under the object's path lock — the gateway's earlier
// checkConditionalHeaders is only a fast-path. Conditions it cannot reduce
// (time-based, combined, ETag lists, weak ETags) keep the pre-change
// semantics: check and commit run under the distributed write lock, which
// routed writes still bypass.
// checkConditionalHeaders is only a fast-path. Every header form reduces
// to clauses, so reducible is false only when the headers cannot parse,
// where the lock path reports the same parse error.
cond, reducible := buildWriteCondition(r)
completionBody := func() s3err.ErrorCode {
var prepCode s3err.ErrorCode
+1 -1
View File
@@ -1067,7 +1067,7 @@ func (s3a *S3ApiServer) putToFiler(r *http.Request, filePath string, dataReader
}
// Route the create to the object's owner filer (its per-path lock serializes
// it); conditional/object-lock/non-reducible cases fall back to the lock.
// it); unparseable or unroutable conditions fall back to the lock.
var createCode s3err.ErrorCode
routed := false
routedAmbiguous := false
@@ -287,9 +287,9 @@ func (f *ambiguousPutFiler) ObjectTransaction(_ context.Context, req *filer_pb.O
case filer_pb.WriteCondition_IF_EXISTS:
ok = current != nil
case filer_pb.WriteCondition_IF_ETAG_MATCH:
ok = current != nil && slices.Contains(clause.Etags, md5Hex)
ok = current != nil && etagCandidateMatch(clause.Etags, md5Hex)
case filer_pb.WriteCondition_IF_ETAG_NOT_MATCH:
ok = current == nil || !slices.Contains(clause.Etags, md5Hex)
ok = current == nil || !etagCandidateMatch(clause.Etags, md5Hex)
}
if !ok {
return &filer_pb.ObjectTransactionResponse{ErrorCode: filer_pb.FilerError_PRECONDITION_FAILED}, nil
@@ -317,6 +317,21 @@ func (f *ambiguousPutFiler) ObjectTransaction(_ context.Context, req *filer_pb.O
return &filer_pb.ObjectTransactionResponse{}, f.txnErr
}
// etagCandidateMatch mirrors the filer's strong-comparison canonicalization:
// quotes are stripped on both sides and a weak (W/) candidate never matches.
func etagCandidateMatch(candidates []string, stored string) bool {
for _, c := range candidates {
c = strings.TrimSpace(c)
if strings.HasPrefix(c, "W/") || strings.HasPrefix(c, "w/") {
continue
}
if strings.Trim(c, `"`) == stored {
return true
}
}
return false
}
// ensureParents mirrors the filer's InsertEntry, which creates missing parent
// directory entries for a stored object (e.g. the .versions dir a version file
// lands under).
+33 -26
View File
@@ -78,9 +78,13 @@ func routeWriteCondition(r *http.Request, uniqueWritePath bool) (*filer_pb.Write
}
// buildWriteCondition reduces the request's conditional headers to a
// WriteCondition. ok=false (combined headers, time conditions, ETag lists, weak
// ETags) keeps gateway-side evaluation under the lock; a nil condition with
// ok=true means unconditional.
// WriteCondition evaluated under the owner's object-path lock. Every header
// form reduces to clauses (ETag lists and weak ETags compare through the same
// canonicalization the gateway applies; time conditions become unix bounds),
// composed per RFC 7232 precedence: If-Unmodified-Since is ignored when
// If-Match is present, If-Modified-Since when If-None-Match is. ok=false only
// when the headers cannot even parse; a nil condition with ok=true means
// unconditional.
func buildWriteCondition(r *http.Request) (*filer_pb.WriteCondition, bool) {
headers, errCode := parseConditionalHeaders(r)
if errCode != s3err.ErrNone {
@@ -89,31 +93,24 @@ func buildWriteCondition(r *http.Request) (*filer_pb.WriteCondition, bool) {
if !headers.isSet {
return nil, true
}
if !headers.ifModifiedSince.IsZero() || !headers.ifUnmodifiedSince.IsZero() {
return nil, false
var clauses []*filer_pb.WriteCondition_Clause
if headers.ifMatch == "*" {
clauses = append(clauses, &filer_pb.WriteCondition_Clause{Kind: filer_pb.WriteCondition_IF_EXISTS})
} else if headers.ifMatch != "" {
clauses = append(clauses, &filer_pb.WriteCondition_Clause{Kind: filer_pb.WriteCondition_IF_ETAG_MATCH, Etags: etagList(headers.ifMatch)})
}
hasMatch := headers.ifMatch != ""
hasNoneMatch := headers.ifNoneMatch != ""
switch {
case hasMatch && !hasNoneMatch:
if headers.ifMatch == "*" {
return clause(filer_pb.WriteCondition_IF_EXISTS), true
}
if etag, single := singleStrongETag(headers.ifMatch); single {
return etagClause(filer_pb.WriteCondition_IF_ETAG_MATCH, etag), true
}
return nil, false
case hasNoneMatch && !hasMatch:
if headers.ifNoneMatch == "*" {
return clause(filer_pb.WriteCondition_IF_NOT_EXISTS), true
}
if etag, single := singleStrongETag(headers.ifNoneMatch); single {
return etagClause(filer_pb.WriteCondition_IF_ETAG_NOT_MATCH, etag), true
}
return nil, false
default:
return nil, false
if !headers.ifUnmodifiedSince.IsZero() && headers.ifMatch == "" {
clauses = append(clauses, &filer_pb.WriteCondition_Clause{Kind: filer_pb.WriteCondition_IF_UNMODIFIED_SINCE, UnixTime: headers.ifUnmodifiedSince.Unix()})
}
if headers.ifNoneMatch == "*" {
clauses = append(clauses, &filer_pb.WriteCondition_Clause{Kind: filer_pb.WriteCondition_IF_NOT_EXISTS})
} else if headers.ifNoneMatch != "" {
clauses = append(clauses, &filer_pb.WriteCondition_Clause{Kind: filer_pb.WriteCondition_IF_ETAG_NOT_MATCH, Etags: etagList(headers.ifNoneMatch)})
}
if !headers.ifModifiedSince.IsZero() && headers.ifNoneMatch == "" {
clauses = append(clauses, &filer_pb.WriteCondition_Clause{Kind: filer_pb.WriteCondition_IF_MODIFIED_SINCE, UnixTime: headers.ifModifiedSince.Unix()})
}
return &filer_pb.WriteCondition{Clauses: clauses}, true
}
// buildDeleteCondition reduces a DeleteObject's If-Match header to a condition;
@@ -154,6 +151,16 @@ func singleStrongETag(v string) (string, bool) {
return strings.Trim(v, `"`), true
}
// etagList splits a conditional ETag header into its candidates, each kept
// verbatim — the evaluator's canonicalization handles quotes and weak markers.
func etagList(v string) []string {
parts := strings.Split(v, ",")
for i := range parts {
parts[i] = strings.TrimSpace(parts[i])
}
return parts
}
// objectTxnOnFiler applies a routed transaction, preferring the object's owner
// filer but failing over to a live filer when the owner is unreachable, so a
// restarted filer's stale ring address (e.g. a rolled K8s pod's old IP) can't hang
+73 -18
View File
@@ -72,7 +72,7 @@ func TestBuildWriteCondition(t *testing.T) {
t.Fatal("want ok")
}
c := oneClause(t, cond)
if c.Kind != filer_pb.WriteCondition_IF_ETAG_MATCH || len(c.Etags) != 1 || c.Etags[0] != "abc123" {
if c.Kind != filer_pb.WriteCondition_IF_ETAG_MATCH || len(c.Etags) != 1 || c.Etags[0] != `"abc123"` {
t.Fatalf("clause = %+v", c)
}
})
@@ -82,33 +82,84 @@ func TestBuildWriteCondition(t *testing.T) {
t.Fatal("want ok")
}
c := oneClause(t, cond)
if c.Kind != filer_pb.WriteCondition_IF_ETAG_NOT_MATCH || len(c.Etags) != 1 || c.Etags[0] != "abc123" {
if c.Kind != filer_pb.WriteCondition_IF_ETAG_NOT_MATCH || len(c.Etags) != 1 || c.Etags[0] != `"abc123"` {
t.Fatalf("clause = %+v", c)
}
})
t.Run("weak etag falls back", func(t *testing.T) {
if _, ok := buildWriteCondition(reqWith(map[string]string{s3_constants.IfMatch: `W/"abc"`})); ok {
t.Fatal("weak etag must not take the fast path")
t.Run("weak etag reduces verbatim", func(t *testing.T) {
cond, ok := buildWriteCondition(reqWith(map[string]string{s3_constants.IfMatch: `W/"abc"`}))
if !ok {
t.Fatal("want ok")
}
c := oneClause(t, cond)
if c.Kind != filer_pb.WriteCondition_IF_ETAG_MATCH || len(c.Etags) != 1 || c.Etags[0] != `W/"abc"` || c.AllowWeak {
t.Fatalf("clause = %+v", c)
}
})
t.Run("etag list falls back", func(t *testing.T) {
if _, ok := buildWriteCondition(reqWith(map[string]string{s3_constants.IfMatch: `"a","b"`})); ok {
t.Fatal("etag list must not take the fast path")
t.Run("etag list reduces", func(t *testing.T) {
cond, ok := buildWriteCondition(reqWith(map[string]string{s3_constants.IfMatch: `"a", "b"`}))
if !ok {
t.Fatal("want ok")
}
c := oneClause(t, cond)
if c.Kind != filer_pb.WriteCondition_IF_ETAG_MATCH || len(c.Etags) != 2 || c.Etags[0] != `"a"` || c.Etags[1] != `"b"` {
t.Fatalf("clause = %+v", c)
}
})
t.Run("both match and none-match falls back", func(t *testing.T) {
if _, ok := buildWriteCondition(reqWith(map[string]string{
t.Run("both match and none-match reduce to clauses", func(t *testing.T) {
cond, ok := buildWriteCondition(reqWith(map[string]string{
s3_constants.IfMatch: "*",
s3_constants.IfNoneMatch: "*",
})); ok {
t.Fatal("ambiguous combination must not take the fast path")
}))
if !ok {
t.Fatal("want ok")
}
if len(cond.Clauses) != 2 || cond.Clauses[0].Kind != filer_pb.WriteCondition_IF_EXISTS || cond.Clauses[1].Kind != filer_pb.WriteCondition_IF_NOT_EXISTS {
t.Fatalf("clauses = %+v", cond.Clauses)
}
})
t.Run("time-based falls back", func(t *testing.T) {
if _, ok := buildWriteCondition(reqWith(map[string]string{
t.Run("time-based reduces to a unix bound", func(t *testing.T) {
cond, ok := buildWriteCondition(reqWith(map[string]string{
"If-Unmodified-Since": "Wed, 21 Oct 2015 07:28:00 GMT",
}))
if !ok {
t.Fatal("want ok")
}
c := oneClause(t, cond)
want := time.Date(2015, 10, 21, 7, 28, 0, 0, time.UTC).Unix()
if c.Kind != filer_pb.WriteCondition_IF_UNMODIFIED_SINCE || c.UnixTime != want {
t.Fatalf("clause = %+v, want unix %d", c, want)
}
})
t.Run("If-Match hides If-Unmodified-Since", func(t *testing.T) {
cond, ok := buildWriteCondition(reqWith(map[string]string{
s3_constants.IfMatch: `"abc"`,
"If-Unmodified-Since": "Wed, 21 Oct 2015 07:28:00 GMT",
}))
if !ok {
t.Fatal("want ok")
}
if c := oneClause(t, cond); c.Kind != filer_pb.WriteCondition_IF_ETAG_MATCH {
t.Fatalf("clause = %+v", c)
}
})
t.Run("If-None-Match hides If-Modified-Since", func(t *testing.T) {
cond, ok := buildWriteCondition(reqWith(map[string]string{
s3_constants.IfNoneMatch: `"abc"`,
"If-Modified-Since": "Wed, 21 Oct 2015 07:28:00 GMT",
}))
if !ok {
t.Fatal("want ok")
}
if c := oneClause(t, cond); c.Kind != filer_pb.WriteCondition_IF_ETAG_NOT_MATCH {
t.Fatalf("clause = %+v", c)
}
})
t.Run("unparseable date falls back", func(t *testing.T) {
if _, ok := buildWriteCondition(reqWith(map[string]string{
"If-Modified-Since": "not a date",
})); ok {
t.Fatal("time condition must not take the fast path")
t.Fatal("unparseable headers must not reduce")
}
})
}
@@ -637,8 +688,12 @@ func TestRouteWriteCondition(t *testing.T) {
if _, ok := routeWriteCondition(reqWith(map[string]string{s3_constants.IfMatch: `"e"`}), true); ok {
t.Fatal("conditional unique write must not route")
}
// A non-reducible condition bails regardless.
if _, ok := routeWriteCondition(reqWith(map[string]string{s3_constants.IfMatch: `W/"e"`}), false); ok {
t.Fatal("weak etag must not route")
// A weak ETag still routes — the clause carries it verbatim.
if _, ok := routeWriteCondition(reqWith(map[string]string{s3_constants.IfMatch: `W/"e"`}), false); !ok {
t.Fatal("weak etag should route")
}
// An unparseable condition bails regardless.
if _, ok := routeWriteCondition(reqWith(map[string]string{"If-Modified-Since": "not a date"}), false); ok {
t.Fatal("unparseable headers must not route")
}
}
+18 -5
View File
@@ -1,6 +1,7 @@
package weed_server
import (
"fmt"
"strconv"
"strings"
"time"
@@ -159,15 +160,27 @@ func canonicalETag(etag string) (value string, weak bool) {
return strings.Trim(etag, `"`), false
}
// storedEntryETag mirrors the S3 gateway's ETag precedence (the stored
// Seaweed ETag extended attribute, then the chunk/Md5 fallback) so conditional
// comparisons match what the gateway computes, without coupling the filer to
// S3 request handling.
// storedEntryETag mirrors the S3 gateway's ETag precedence — the stored
// Seaweed ETag extended attribute, then the remote ETag while the entry still
// holds the remote bytes, then Md5, then the chunk hash (an empty object has
// no ETag) — so conditional comparisons match what the gateway computes,
// without coupling the filer to S3 request handling.
func storedEntryETag(entry *filer.Entry) string {
if v, ok := entry.Extended[s3_constants.ExtETagKey]; ok && len(v) > 0 {
return normalizeETag(string(v))
}
return normalizeETag(filer.ETagEntry(entry))
if remote := entry.Remote; remote != nil && remote.RemoteETag != "" &&
entry.FileSize == uint64(remote.RemoteSize) &&
(remote.LastLocalSyncTsNs > 0 || (len(entry.GetChunks()) == 0 && len(entry.Content) == 0)) {
return normalizeETag(remote.RemoteETag)
}
if len(entry.Md5) > 0 {
return fmt.Sprintf("%x", entry.Md5)
}
if len(entry.GetChunks()) == 0 {
return ""
}
return normalizeETag(filer.ETagChunks(entry.GetChunks()))
}
func normalizeETag(etag string) string {
@@ -257,6 +257,23 @@ func TestStoredEntryETag(t *testing.T) {
if got := storedEntryETag(md5Only); got != "abcd" {
t.Errorf("md5 fallback: got %q", got)
}
remoteOnly := &filer.Entry{
Attr: filer.Attr{FileSize: 10},
Remote: &filer_pb.RemoteEntry{RemoteETag: `"remote-etag"`, RemoteSize: 10},
}
if got := storedEntryETag(remoteOnly); got != "remote-etag" {
t.Errorf("remote etag: got %q", got)
}
remoteStale := &filer.Entry{
Attr: filer.Attr{FileSize: 10, Md5: []byte{0xab}},
Remote: &filer_pb.RemoteEntry{RemoteETag: `"remote-etag"`, RemoteSize: 99},
}
if got := storedEntryETag(remoteStale); got != "ab" {
t.Errorf("stale remote size falls to md5: got %q", got)
}
if got := storedEntryETag(&filer.Entry{}); got != "" {
t.Errorf("empty entry: got %q", got)
}
}
// The CreateEntry handler enforces the precondition atomically: a matching