Compare commits

..
Author SHA1 Message Date
Chris Lu 7bf30b5801 s3: report an empty probed directory as a common prefix, not a key
The "<dir>/" existence probe emitted the empty directory as a
FolderMimeType Contents entry, so delimited listings returned a fake
readable key that breaks clients treating every key as an object.
Delimited probes now report the directory as a CommonPrefix -- which
the probe accepts for existence -- while the recursive (no delimiter)
probe keeps the directory marker.
2026-10-16 00:25:21 -07:00
8 changed files with 69 additions and 19 deletions
+2 -2
View File
@@ -114,7 +114,7 @@ func TestS3Integration(t *testing.T) {
// findAvailablePort finds an available port by binding to port 0
// startMiniCluster starts a weed mini instance directly without exec.
// Extra flags can be appended via extraArgs.
// Extra flags (e.g. "-s3.allowDeleteBucketNotEmpty=false") can be appended via extraArgs.
func startMiniCluster(t *testing.T, extraArgs ...string) (*TestCluster, error) {
// Allocate non-colliding ports (including gRPC offsets) for weed mini.
ports := testutil.MustFreeMiniPorts(t, []string{"Master", "Volume", "Filer", "S3"})
@@ -872,7 +872,7 @@ func TestS3DeleteBucketNotEmpty(t *testing.T) {
t.Skip("Skipping integration test in short mode")
}
cluster, err := startMiniCluster(t)
cluster, err := startMiniCluster(t, "-s3.allowDeleteBucketNotEmpty=false")
require.NoError(t, err)
defer cluster.Stop()
+2 -2
View File
@@ -228,8 +228,8 @@ test-with-server: start-server
# Test with different configurations
test-versioning-with-configs: check-deps
@echo "Testing with different S3 configurations..."
@echo "Testing with delete bucket not empty allowed..."
@$(WEED_BINARY) mini -s3.port=$(S3_PORT) -s3.allowDeleteBucketNotEmpty=true > weed-test-config1.log 2>&1 & echo $$! > weed-config1.pid
@echo "Testing with empty folder allowed..."
@$(WEED_BINARY) mini -s3.port=$(S3_PORT) > weed-test-config1.log 2>&1 & echo $$! > weed-config1.pid
@sleep 5
@go test -v -timeout=5m -run "TestVersioningBasicWorkflow" . || true
@if [ -f weed-config1.pid ]; then kill -TERM $$(cat weed-config1.pid) 2>/dev/null || true; rm -f weed-config1.pid; fi
+1 -1
View File
@@ -154,7 +154,7 @@ func init() {
filerS3Options.metricsHttpPort = cmdFiler.Flag.Int("s3.metricsPort", 0, "Prometheus metrics listen port")
filerS3Options.metricsHttpIp = cmdFiler.Flag.String("s3.metricsIp", "", "metrics listen ip. If empty, default to same as -s3.ip.bind option.")
cmdFiler.Flag.Bool("s3.allowEmptyFolder", true, "deprecated, ignored. Empty folder cleanup is now automatic.")
filerS3Options.allowDeleteBucketNotEmpty = cmdFiler.Flag.Bool("s3.allowDeleteBucketNotEmpty", false, "allow recursive deleting all entries along with bucket")
filerS3Options.allowDeleteBucketNotEmpty = cmdFiler.Flag.Bool("s3.allowDeleteBucketNotEmpty", true, "allow recursive deleting all entries along with bucket")
filerS3Options.autoCreateBucket = cmdFiler.Flag.Bool("s3.autoCreateBucket", true, "create the bucket on upload if it does not exist, for admin identities only")
filerS3Options.localSocket = cmdFiler.Flag.String("s3.localSocket", "", "default to /tmp/seaweedfs-s3-<port>.sock")
filerS3Options.tlsCACertificate = cmdFiler.Flag.String("s3.cacert.file", "", "path to the TLS CA certificate file")
+1 -1
View File
@@ -407,7 +407,7 @@ var (
miniMetricsHttpIp = cmdMini.Flag.String("metricsIp", "", "metrics listen ip. If empty, default to same as -ip.bind option.")
miniS3Config = cmdMini.Flag.String("s3.config", "", "path to the S3 config file")
miniIamConfig = cmdMini.Flag.String("s3.iam.config", "", "path to the advanced IAM config file for S3")
miniS3AllowDeleteBucketNotEmpty = cmdMini.Flag.Bool("s3.allowDeleteBucketNotEmpty", false, "allow recursive deleting all entries along with bucket")
miniS3AllowDeleteBucketNotEmpty = cmdMini.Flag.Bool("s3.allowDeleteBucketNotEmpty", true, "allow recursive deleting all entries along with bucket")
miniS3AutoCreateBucket = cmdMini.Flag.Bool("s3.autoCreateBucket", true, "create the bucket on upload if it does not exist, for admin identities only")
miniBucket = cmdMini.Flag.String("bucket", "", "comma-separated S3 bucket names to create on startup if they do not already exist; leave empty to skip. Falls back to S3_BUCKET env var.")
miniTableBucket = cmdMini.Flag.String("tableBucket", "", "comma-separated S3 Tables buckets to create on startup if they do not already exist, each name[:FORMAT] with FORMAT one of ICEBERG (default) or LANCE, e.g. warehouse,vectors:LANCE; leave empty to skip. Falls back to S3_TABLE_BUCKET env var.")
+1 -1
View File
@@ -114,7 +114,7 @@ func init() {
s3StandaloneOptions.metricsHttpPort = cmdS3.Flag.Int("metricsPort", 0, "Prometheus metrics listen port")
s3StandaloneOptions.metricsHttpIp = cmdS3.Flag.String("metricsIp", "", "metrics listen ip. If empty, default to same as -ip.bind option.")
cmdS3.Flag.Bool("allowEmptyFolder", true, "deprecated, ignored. Empty folder cleanup is now automatic.")
s3StandaloneOptions.allowDeleteBucketNotEmpty = cmdS3.Flag.Bool("allowDeleteBucketNotEmpty", false, "allow recursive deleting all entries along with bucket")
s3StandaloneOptions.allowDeleteBucketNotEmpty = cmdS3.Flag.Bool("allowDeleteBucketNotEmpty", true, "allow recursive deleting all entries along with bucket")
s3StandaloneOptions.autoCreateBucket = cmdS3.Flag.Bool("autoCreateBucket", true, "create the bucket on upload if it does not exist, for admin identities only")
s3StandaloneOptions.localFilerSocket = cmdS3.Flag.String("localFilerSocket", "", "local filer socket path")
s3StandaloneOptions.localSocket = cmdS3.Flag.String("localSocket", "", "default to /tmp/seaweedfs-s3-<port>.sock")
+1 -1
View File
@@ -182,7 +182,7 @@ func init() {
s3Options.iamConfig = cmdServer.Flag.String("s3.iam.config", "", "path to the advanced IAM config file for S3. Overrides -iam.config if both are provided.")
s3Options.auditLogConfig = cmdServer.Flag.String("s3.auditLogConfig", "", "path to the audit log config file")
cmdServer.Flag.Bool("s3.allowEmptyFolder", true, "deprecated, ignored. Empty folder cleanup is now automatic.")
s3Options.allowDeleteBucketNotEmpty = cmdServer.Flag.Bool("s3.allowDeleteBucketNotEmpty", false, "allow recursive deleting all entries along with bucket")
s3Options.allowDeleteBucketNotEmpty = cmdServer.Flag.Bool("s3.allowDeleteBucketNotEmpty", true, "allow recursive deleting all entries along with bucket")
s3Options.autoCreateBucket = cmdServer.Flag.Bool("s3.autoCreateBucket", true, "create the bucket on upload if it does not exist, for admin identities only")
s3Options.localSocket = cmdServer.Flag.String("s3.localSocket", "", "default to /tmp/seaweedfs-s3-<port>.sock")
s3Options.bindIp = cmdServer.Flag.String("s3.ip.bind", "", "ip address to bind to. If empty, default to same as -ip.bind option.")
+26 -5
View File
@@ -539,6 +539,18 @@ func (s3a *S3ApiServer) listFilerEntries(ctx context.Context, req listObjectsReq
}
dirName, entryName, _ := entryUrlEncode(dir, entry.Name, encodingTypeUrl)
if entry.IsDirectory {
if cursor.emitDirAsCommonPrefix {
cursor.emitDirAsCommonPrefix = false
dirPrefix := fmt.Sprintf("%s/%s/", dir, entry.Name)[len(bucketPrefix):]
commonPrefixes = append(commonPrefixes, PrefixEntry{
Prefix: dirPrefix,
})
cursor.maxKeys--
lastEntryWasCommonPrefix = true
lastCommonPrefix = dirPrefix
lastPrefixNullBackers, lastPrefixConfirmed = nil, true
return
}
if originalPrefix != "" {
normalizedPrefix := strings.TrimPrefix(strings.TrimSuffix(originalPrefix, "/"), "/")
if normalizedPrefix != "" {
@@ -789,6 +801,9 @@ type ListingCursor struct {
// resolvePendingNulls settles trailing null objects whose .versions sibling
// has not streamed yet before a page is declared full.
resolvePendingNulls func() error
// emitDirAsCommonPrefix marks the next directory entry passed to the entry
// callback for reporting as a CommonPrefix rather than a Contents key.
emitDirAsCommonPrefix bool
}
// excludedMarkerKey returns the key an exclusive marker names that the walk's cutoff no
@@ -1124,15 +1139,21 @@ func (s3a *S3ApiServer) doListFilerEntries(ctx context.Context, client filer_pb.
// A real but empty directory (created out of band via mount, mkdir or
// the filer API, so it carries no MIME) is otherwise invisible to S3
// clients that detect directories by listing it under its own "<dir>/"
// prefix. Surface it as a directory marker for that explicit probe,
// identical to a directory created via PutObject with a trailing "/", so
// tools like hadoop-aws can find it. Plain listings are left untouched, so
// empty directories left behind by deleted objects are not shown as keys.
// prefix. Answer that explicit probe so tools like hadoop-aws can find
// the directory: a delimited listing reports it as a CommonPrefix --
// a real key in Contents would look like an object a client can read --
// while a recursive listing keeps the directory marker key it used.
// Plain listings are left untouched, so empty directories left behind
// by deleted objects are not shown as keys.
// A directory that still holds version history no longer names anything,
// so it gets no marker either.
if explicitDirProbe && !isKeyObject && !childEmitted && !cursor.isTruncated && entry.Attributes != nil &&
!s3a.dirHoldsOnlyHiddenEntries(ctx, client, bucket, dir+"/"+entry.Name, cursor) {
entry.Attributes.Mime = s3_constants.FolderMimeType
if delimiter == "/" {
cursor.emitDirAsCommonPrefix = true
} else {
entry.Attributes.Mime = s3_constants.FolderMimeType
}
eachEntryFn(dir, entry)
}
// println("doListFilerEntries2 dir", dir+"/"+entry.Name, "subNextMarker", subNextMarker)
@@ -3,10 +3,11 @@ package s3api
import (
"context"
"testing"
"time"
"github.com/seaweedfs/seaweedfs/weed/pb/filer_pb"
"github.com/seaweedfs/seaweedfs/weed/s3api/s3_constants"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
)
// TestDirectoryListedAsCommonPrefix tests that regular directories (without MIME)
@@ -67,8 +68,8 @@ func TestDirectoryListedAsCommonPrefix(t *testing.T) {
// TestEmptyDirectorySurfacedAsMarker reproduces the hadoop-aws / Spark case where a
// real but empty directory (created via mount, mkdir or the filer API, so it has no
// MIME) must be visible to S3 clients that detect directories by listing under the
// "<dir>/" prefix. Such a directory is surfaced as a directory marker, identical to
// one created via PutObject with a trailing "/".
// "<dir>/" prefix. Under a delimiter it is reported as a CommonPrefix, which the
// probe accepts, instead of a key a client might try to read.
func TestEmptyDirectorySurfacedAsMarker(t *testing.T) {
s3a := &S3ApiServer{
option: &S3ApiServerOption{BucketsPath: "/buckets"},
@@ -96,9 +97,10 @@ func TestEmptyDirectorySurfacedAsMarker(t *testing.T) {
assert.NoError(t, err)
if assert.Len(t, seen, 1, "empty directory must be surfaced under the logs/ prefix") {
assert.True(t, seen[0].IsDirectoryKeyObject(),
"empty directory must be marked as a directory key object so it lists as logs/")
assert.Equal(t, s3_constants.FolderMimeType, seen[0].Attributes.Mime)
assert.True(t, cursor.emitDirAsCommonPrefix,
"a delimited probe reports the empty directory as a common prefix")
assert.Empty(t, seen[0].Attributes.Mime,
"the entry must not be dressed as a directory key object")
}
}
@@ -182,3 +184,30 @@ func TestEmptyDirectoryHiddenInFlatListing(t *testing.T) {
assert.NoError(t, err)
assert.Equal(t, []string{"foo"}, seen, "flat listing must hide the empty directory and return only real objects")
}
// A trailing-slash delimited listing of an empty directory reports the
// directory as a CommonPrefix, not as a Contents key: clients that treat
// every returned key as an object (e.g. lock listings) skip prefixes.
func TestEmptyDirectoryListedAsCommonPrefix(t *testing.T) {
filer := &fakeBucketDeleteFiler{entriesByDir: map[string][]*filer_pb.Entry{
"/buckets/test": {
{Name: "locks", IsDirectory: true, Attributes: &filer_pb.FuseAttributes{Mime: ""}},
},
"/buckets/test/locks": {},
}}
s3a := newFailoverTestServer(t, startFakeFiler(t, filer))
s3a.option.BucketsPath = "/buckets"
s3a.bucketConfigCache = NewBucketConfigCache(time.Minute)
s3a.bucketConfigCache.Set("test", &BucketConfig{Name: "test"})
response, err := s3a.listFilerEntries(context.Background(), listObjectsRequest{
bucket: "test",
prefix: "locks/",
delimiter: "/",
maxKeys: 1000,
})
require.NoError(t, err)
assert.Empty(t, response.Contents, "an empty directory is not a readable key")
require.Len(t, response.CommonPrefixes, 1)
assert.Equal(t, "locks/", response.CommonPrefixes[0].Prefix)
}