feat: drop table location mapping support (#8458)

* feat: drop table location mapping support

Disable external metadata locations for S3 Tables and remove the table location
mapping index entirely. Table metadata must live under the table bucket paths,
so lookups no longer use mapping directories.

Changes:
- Remove mapping lookup and cache from bucket path resolution
- Reject metadataLocation in CreateTable and UpdateTable
- Remove mapping helpers and tests

* compile

* refactor

* fix: accept metadataLocation in S3 Tables API requests

We removed the external table location mapping feature, but still need to
accept and store metadataLocation values from clients like Trino. The mapping
feature was an internal implementation detail that mapped external buckets to
internal table paths. The metadataLocation field itself is part of the S3 Tables
API and should be preserved.

* fmt

* fix: handle MetadataLocation in UpdateTable requests

Mirror handleCreateTable behavior by updating metadata.MetadataLocation
when req.MetadataLocation is provided in UpdateTable requests. This ensures
table metadata location can be updated, not just set during creation.
This commit is contained in:
Chris Lu
2026-02-26 16:36:24 -08:00
committed by GitHub
parent 641351da78
commit 8eba7ba5b2
8 changed files with 50 additions and 483 deletions

View File

@@ -1,7 +1,6 @@
package s3tables
import (
"context"
"encoding/json"
"errors"
"fmt"
@@ -10,7 +9,6 @@ import (
"strings"
"time"
"github.com/seaweedfs/seaweedfs/weed/glog"
"github.com/seaweedfs/seaweedfs/weed/pb/filer_pb"
)
@@ -242,10 +240,6 @@ func (h *S3TablesHandler) handleCreateTable(w http.ResponseWriter, r *http.Reque
}
}
if err := h.updateTableLocationMapping(r.Context(), client, "", req.MetadataLocation, tablePath); err != nil {
glog.V(1).Infof("failed to update table location mapping for %s: %v", req.MetadataLocation, err)
}
return nil
})
@@ -943,9 +937,6 @@ func (h *S3TablesHandler) handleDeleteTable(w http.ResponseWriter, r *http.Reque
if err := h.deleteDirectory(r.Context(), client, tablePath); err != nil {
return err
}
if err := h.deleteTableLocationMapping(r.Context(), client, metadata.MetadataLocation, tablePath); err != nil {
glog.V(1).Infof("failed to delete table location mapping for %s: %v", metadata.MetadataLocation, err)
}
return nil
})
@@ -1090,9 +1081,6 @@ func (h *S3TablesHandler) handleUpdateTable(w http.ResponseWriter, r *http.Reque
return ErrVersionTokenMismatch
}
// Capture old metadata location before mutation for stale mapping cleanup
oldMetadataLocation := metadata.MetadataLocation
// Update metadata
if req.Metadata != nil {
if metadata.Metadata == nil {
@@ -1131,9 +1119,6 @@ func (h *S3TablesHandler) handleUpdateTable(w http.ResponseWriter, r *http.Reque
if err := h.setExtendedAttribute(r.Context(), client, tablePath, ExtendedKeyMetadata, metadataBytes); err != nil {
return err
}
if err := h.updateTableLocationMapping(r.Context(), client, oldMetadataLocation, metadata.MetadataLocation, tablePath); err != nil {
glog.V(1).Infof("failed to update table location mapping for %s -> %s: %v", oldMetadataLocation, metadata.MetadataLocation, err)
}
return nil
})
@@ -1149,104 +1134,3 @@ func (h *S3TablesHandler) handleUpdateTable(w http.ResponseWriter, r *http.Reque
})
return nil
}
func (h *S3TablesHandler) updateTableLocationMapping(ctx context.Context, client filer_pb.SeaweedFilerClient, oldMetadataLocation, newMetadataLocation, tablePath string) error {
newTableLocationBucket, ok := parseTableLocationBucket(newMetadataLocation)
if !ok {
return nil
}
tableBucketPath, ok := tableBucketPathFromTablePath(tablePath)
if !ok {
return fmt.Errorf("invalid table path for location mapping: %s", tablePath)
}
if err := h.ensureDirectory(ctx, client, GetTableLocationMappingDir()); err != nil {
return err
}
if err := h.ensureTableLocationMappingBucketDir(ctx, client, newTableLocationBucket); err != nil {
return err
}
// If the metadata location changed, remove this table's stale mapping entry from the old bucket.
if oldMetadataLocation != "" && oldMetadataLocation != newMetadataLocation {
oldTableLocationBucket, ok := parseTableLocationBucket(oldMetadataLocation)
if ok && oldTableLocationBucket != newTableLocationBucket {
if err := h.removeTableLocationMappingEntry(ctx, client, oldTableLocationBucket, tablePath); err != nil {
glog.V(1).Infof("failed to delete stale mapping for %s: %v", oldTableLocationBucket, err)
}
}
}
return h.upsertFile(ctx, client, GetTableLocationMappingEntryPath(newTableLocationBucket, tablePath), []byte(tableBucketPath))
}
func (h *S3TablesHandler) deleteTableLocationMapping(ctx context.Context, client filer_pb.SeaweedFilerClient, metadataLocation, tablePath string) error {
tableLocationBucket, ok := parseTableLocationBucket(metadataLocation)
if !ok {
return nil
}
return h.removeTableLocationMappingEntry(ctx, client, tableLocationBucket, tablePath)
}
func (h *S3TablesHandler) ensureTableLocationMappingBucketDir(ctx context.Context, client filer_pb.SeaweedFilerClient, tableLocationBucket string) error {
mappingDir := GetTableLocationMappingDir()
bucketMappingPath := GetTableLocationMappingPath(tableLocationBucket)
resp, err := filer_pb.LookupEntry(ctx, client, &filer_pb.LookupDirectoryEntryRequest{
Directory: mappingDir,
Name: tableLocationBucket,
})
if err == nil {
if resp != nil && resp.Entry != nil && resp.Entry.IsDirectory {
return nil
}
if removeErr := h.deleteEntryIfExists(ctx, client, bucketMappingPath); removeErr != nil && !errors.Is(removeErr, filer_pb.ErrNotFound) {
return removeErr
}
} else if !errors.Is(err, filer_pb.ErrNotFound) {
return err
}
return h.ensureDirectory(ctx, client, bucketMappingPath)
}
func (h *S3TablesHandler) removeTableLocationMappingEntry(ctx context.Context, client filer_pb.SeaweedFilerClient, tableLocationBucket, tablePath string) error {
entryPath := GetTableLocationMappingEntryPath(tableLocationBucket, tablePath)
if err := h.deleteEntryIfExists(ctx, client, entryPath); err != nil && !errors.Is(err, filer_pb.ErrNotFound) {
return err
}
return h.removeTableLocationMappingBucketDirIfEmpty(ctx, client, tableLocationBucket)
}
func (h *S3TablesHandler) removeTableLocationMappingBucketDirIfEmpty(ctx context.Context, client filer_pb.SeaweedFilerClient, tableLocationBucket string) error {
bucketMappingPath := GetTableLocationMappingPath(tableLocationBucket)
stream, err := client.ListEntries(ctx, &filer_pb.ListEntriesRequest{
Directory: bucketMappingPath,
Limit: 1,
})
if err != nil {
if errors.Is(err, filer_pb.ErrNotFound) {
return nil
}
return err
}
for {
resp, recvErr := stream.Recv()
if recvErr == io.EOF {
break
}
if recvErr != nil {
return recvErr
}
if resp != nil && resp.Entry != nil {
return nil
}
}
if err := h.deleteEntryIfExists(ctx, client, bucketMappingPath); err != nil && !errors.Is(err, filer_pb.ErrNotFound) {
return err
}
return nil
}

View File

@@ -1,77 +0,0 @@
package s3tables
import (
"strings"
"testing"
)
func TestGetTableLocationMappingEntryPathPerTable(t *testing.T) {
tableLocationBucket := "shared-location--table-s3"
tablePathA := GetTablePath("warehouse", "analytics", "orders")
tablePathB := GetTablePath("warehouse", "analytics", "customers")
entryPathA := GetTableLocationMappingEntryPath(tableLocationBucket, tablePathA)
entryPathARepeat := GetTableLocationMappingEntryPath(tableLocationBucket, tablePathA)
entryPathB := GetTableLocationMappingEntryPath(tableLocationBucket, tablePathB)
if entryPathA != entryPathARepeat {
t.Fatalf("mapping entry path should be deterministic: %q != %q", entryPathA, entryPathARepeat)
}
if entryPathA == entryPathB {
t.Fatalf("mapping entry path should differ per table path: %q == %q", entryPathA, entryPathB)
}
expectedPrefix := GetTableLocationMappingPath(tableLocationBucket) + "/"
if !strings.HasPrefix(entryPathA, expectedPrefix) {
t.Fatalf("mapping entry path %q should start with %q", entryPathA, expectedPrefix)
}
if strings.TrimPrefix(entryPathA, expectedPrefix) == "" {
t.Fatalf("mapping entry name should not be empty: %q", entryPathA)
}
}
func TestTableBucketPathFromTablePath(t *testing.T) {
testCases := []struct {
name string
tablePath string
expected string
ok bool
}{
{
name: "valid table path",
tablePath: GetTablePath("warehouse", "analytics", "orders"),
expected: GetTableBucketPath("warehouse"),
ok: true,
},
{
name: "valid table bucket root",
tablePath: GetTableBucketPath("warehouse"),
expected: GetTableBucketPath("warehouse"),
ok: true,
},
{
name: "invalid non-tables path",
tablePath: "/tmp/warehouse/analytics/orders",
expected: "",
ok: false,
},
{
name: "invalid empty bucket segment",
tablePath: "/buckets/",
expected: "",
ok: false,
},
}
for _, tc := range testCases {
t.Run(tc.name, func(t *testing.T) {
actual, ok := tableBucketPathFromTablePath(tc.tablePath)
if ok != tc.ok {
t.Fatalf("tableBucketPathFromTablePath(%q) ok=%v, want %v", tc.tablePath, ok, tc.ok)
}
if actual != tc.expected {
t.Fatalf("tableBucketPathFromTablePath(%q)=%q, want %q", tc.tablePath, actual, tc.expected)
}
})
}
}

View File

@@ -2,7 +2,6 @@ package s3tables
import (
"crypto/rand"
"crypto/sha1"
"encoding/hex"
"fmt"
"net/url"
@@ -21,8 +20,7 @@ const (
)
const (
tableLocationMappingsDirPath = "/etc/s3tables"
tableObjectRootDirName = ".objects"
tableObjectRootDirName = ".objects"
)
var (
@@ -109,43 +107,6 @@ func GetTableObjectBucketPath(bucketName string) string {
return path.Join(GetTableObjectRootDir(), bucketName)
}
// GetTableLocationMappingDir returns the root path for table location bucket mappings
func GetTableLocationMappingDir() string {
return tableLocationMappingsDirPath
}
// GetTableLocationMappingPath returns the filer path for a table location bucket mapping
func GetTableLocationMappingPath(tableLocationBucket string) string {
return path.Join(GetTableLocationMappingDir(), tableLocationBucket)
}
// GetTableLocationMappingEntryPath returns the filer path for a table-specific mapping entry.
// Each table gets its own entry so multiple tables can share the same external table-location bucket.
func GetTableLocationMappingEntryPath(tableLocationBucket, tablePath string) string {
return path.Join(GetTableLocationMappingPath(tableLocationBucket), tableLocationMappingEntryName(tablePath))
}
func tableLocationMappingEntryName(tablePath string) string {
normalized := path.Clean("/" + strings.TrimSpace(strings.TrimPrefix(tablePath, "/")))
sum := sha1.Sum([]byte(normalized))
return hex.EncodeToString(sum[:])
}
func tableBucketPathFromTablePath(tablePath string) (string, bool) {
normalized := path.Clean("/" + strings.TrimSpace(strings.TrimPrefix(tablePath, "/")))
tablesPrefix := strings.TrimSuffix(TablesPath, "/") + "/"
if !strings.HasPrefix(normalized, tablesPrefix) {
return "", false
}
remaining := strings.TrimPrefix(normalized, tablesPrefix)
bucketName, _, _ := strings.Cut(remaining, "/")
if bucketName == "" {
return "", false
}
return path.Join(TablesPath, bucketName), true
}
// Metadata structures
type tableBucketMetadata struct {
@@ -244,22 +205,6 @@ func ValidateBucketName(name string) error {
return validateBucketName(name)
}
func parseTableLocationBucket(metadataLocation string) (string, bool) {
if !strings.HasPrefix(metadataLocation, "s3://") {
return "", false
}
trimmed := strings.TrimPrefix(metadataLocation, "s3://")
trimmed = strings.TrimSuffix(trimmed, "/")
if trimmed == "" {
return "", false
}
bucket, _, _ := strings.Cut(trimmed, "/")
if bucket == "" || !strings.HasSuffix(bucket, "--table-s3") {
return "", false
}
return bucket, true
}
// BuildBucketARN builds a bucket ARN with the provided region and account ID.
// If region is empty, the ARN will omit the region field.
func BuildBucketARN(region, accountID, bucketName string) (string, error) {