Files
seaweedfs/weed/s3api/s3tables/handler_bucket_create.go
T
Chris Lu a724f4296c s3, s3tables: create buckets and namespaces exclusively (#11872)
* filer_pb: add MkdirExclusive

A taken name leaves the callback as a result, not an error, so filer
failover does not count a lost race as a filer failure.

* s3tables: create the entry with its attributes in one exclusive create

CreateTableBucket and CreateNamespace checked that the name was free and
then wrote the entry in several calls, so a second create in the window
replaced the first caller's entry including its owner. The create now
carries the marker, metadata and tags in one O_EXCL CreateEntry that
answers ErrEntryAlreadyExists with the same 409.

* s3: PutBucket uses an exclusive mkdir

The same window let a plain PutBucket replace a table bucket created
concurrently. The already-exists branch now answers from the entry itself,
since the registry just cached the name as absent, and the admin
auto-create path is exclusive too.

* s3: read the winning entry when a create loses the race

The concurrent-create branch called existingBucketError on a name the
registry had just cached as absent, answering NoSuchBucket, and the
auto-create fallback could assign an S3 owner to a table bucket it met
there. The negative cache is cleared before the existing-bucket error is
built, and a table bucket keeps its owner-less shape.

* s3: fail an auto-create that met a table bucket

A table bucket carries no S3 owner and must not gain one, and the upload
that auto-create ran for must not continue into it either.

* s3: fail an auto-create when the winning entry cannot be read

An unreadable existing entry fell through to success, so the write
continued into an unvalidated entry, possibly a table bucket.
2026-10-20 19:01:08 -07:00

194 lines
6.3 KiB
Go

package s3tables
import (
"encoding/json"
"errors"
"fmt"
"net/http"
"time"
"github.com/seaweedfs/seaweedfs/weed/glog"
"github.com/seaweedfs/seaweedfs/weed/pb/filer_pb"
"github.com/seaweedfs/seaweedfs/weed/s3api/s3_constants"
)
// handleCreateTableBucket creates a new table bucket
func (h *S3TablesHandler) handleCreateTableBucket(w http.ResponseWriter, r *http.Request, filerClient FilerClient) error {
var req CreateTableBucketRequest
if err := h.readRequestBody(r, &req); err != nil {
h.writeError(w, http.StatusBadRequest, ErrCodeInvalidRequest, err.Error())
return err
}
// Validate bucket name
if err := validateBucketName(req.Name); err != nil {
h.writeError(w, http.StatusBadRequest, ErrCodeInvalidRequest, err.Error())
return err
}
// A bucket is a catalog, and a catalog serves one protocol. Saying which one
// at creation is what lets everything downstream - the endpoint the UI
// shows, the tables the bucket accepts - be answered without opening a table.
bucketFormat := FormatIceberg
if req.Format != "" {
normalized, ok := NormalizeFormat(req.Format)
if !ok {
h.writeError(w, http.StatusBadRequest, ErrCodeInvalidRequest, fmt.Sprintf("unsupported format %q", req.Format))
return fmt.Errorf("invalid format")
}
bucketFormat = normalized
}
principal := h.getAccountID(r)
identityActions := getIdentityActions(r)
identityPolicyNames := getIdentityPolicyNames(r)
useIAM := h.shouldUseIAM(r, identityActions, identityPolicyNames)
useLegacy := !useIAM
if useIAM {
allowed, err := h.authorizeIAMAction(r, identityPolicyNames, "CreateTableBucket", h.generateTableBucketARN(principal, req.Name), h.generateS3BucketARN(req.Name))
if err != nil {
h.writeError(w, http.StatusForbidden, ErrCodeAccessDenied, "not authorized to create table buckets")
return NewAuthError("CreateTableBucket", principal, "not authorized to create table buckets")
}
if !allowed {
if h.defaultAllow && len(identityActions) == 0 && len(identityPolicyNames) == 0 {
useLegacy = true
} else {
h.writeError(w, http.StatusForbidden, ErrCodeAccessDenied, "not authorized to create table buckets")
return NewAuthError("CreateTableBucket", principal, "not authorized to create table buckets")
}
}
}
if useLegacy {
owner := h.accountID
if owner == "" {
owner = DefaultAccountID
}
if !CheckPermissionWithContext("CreateTableBucket", principal, owner, "", "", &PolicyContext{
TrustedIssuers: h.trustedIssuers(r),
IsLocalAccountID: h.localAccountIDs,
IdentityActions: identityActions,
DefaultAllow: h.defaultAllowFor(r),
}) {
h.writeError(w, http.StatusForbidden, ErrCodeAccessDenied, "not authorized to create table buckets")
return NewAuthError("CreateTableBucket", principal, "not authorized to create table buckets")
}
}
bucketPath := GetTableBucketPath(req.Name)
// Check if bucket already exists and ensure no conflict with object store buckets
tableBucketExists := false
s3BucketExists := false
bucketsPath := s3_constants.DefaultBucketsPath
err := filerClient.WithFilerClient(false, func(client filer_pb.SeaweedFilerClient) error {
resp, err := client.GetFilerConfiguration(r.Context(), &filer_pb.GetFilerConfigurationRequest{})
if err != nil {
return err
}
if resp.DirBuckets != "" {
bucketsPath = resp.DirBuckets
}
entryResp, err := filer_pb.LookupEntry(r.Context(), client, &filer_pb.LookupDirectoryEntryRequest{
Directory: bucketsPath,
Name: req.Name,
})
if err != nil {
if !errors.Is(err, filer_pb.ErrNotFound) {
return err
}
return nil
}
if entryResp != nil && entryResp.Entry != nil {
if IsTableBucketEntry(entryResp.Entry) {
tableBucketExists = true
} else {
s3BucketExists = true
}
}
return nil
})
if err != nil {
glog.Errorf("S3Tables: failed to check bucket existence: %v", err)
h.writeError(w, http.StatusInternalServerError, ErrCodeInternalError, "failed to check bucket existence")
return err
}
if tableBucketExists {
h.writeError(w, http.StatusConflict, ErrCodeBucketAlreadyExists, fmt.Sprintf("table bucket %s already exists", req.Name))
return fmt.Errorf("bucket already exists")
}
if s3BucketExists {
h.writeError(w, http.StatusConflict, ErrCodeBucketAlreadyExists, fmt.Sprintf("bucket %s already exists", req.Name))
return fmt.Errorf("bucket already exists")
}
// Create the bucket directory and set metadata as extended attributes
now := time.Now()
metadata := &tableBucketMetadata{
Name: req.Name,
CreatedAt: now,
OwnerAccountID: principal,
Format: bucketFormat,
}
metadataBytes, err := json.Marshal(metadata)
if err != nil {
glog.Errorf("S3Tables: failed to marshal metadata: %v", err)
h.writeError(w, http.StatusInternalServerError, ErrCodeInternalError, "failed to marshal metadata")
return err
}
var taken error
err = filerClient.WithFilerClient(false, func(client filer_pb.SeaweedFilerClient) error {
// Ensure root tables directory exists
if !h.entryExists(r.Context(), client, TablesPath) {
if err := h.createDirectory(r.Context(), client, TablesPath); err != nil {
return fmt.Errorf("failed to create root tables directory: %w", err)
}
}
extended := map[string][]byte{
ExtendedKeyTableBucket: []byte("true"),
ExtendedKeyMetadata: metadataBytes,
}
if len(req.Tags) > 0 {
tagsBytes, err := json.Marshal(req.Tags)
if err != nil {
return fmt.Errorf("failed to marshal tags: %w", err)
}
extended[ExtendedKeyTags] = tagsBytes
}
if err := h.createDirectoryExclusive(r.Context(), client, bucketPath, extended); err != nil {
if errors.Is(err, filer_pb.ErrEntryAlreadyExists) {
taken = err // an answer, not a filer failure
return nil
}
return err
}
return nil
})
if err == nil && taken != nil {
err = taken
}
if errors.Is(err, filer_pb.ErrEntryAlreadyExists) {
h.writeError(w, http.StatusConflict, ErrCodeBucketAlreadyExists, fmt.Sprintf("table bucket %s already exists", req.Name))
return err
}
if err != nil {
glog.Errorf("S3Tables: failed to create table bucket %s: %v", req.Name, err)
h.writeError(w, http.StatusInternalServerError, ErrCodeInternalError, "failed to create table bucket")
return err
}
resp := &CreateTableBucketResponse{
ARN: h.generateTableBucketARN(metadata.OwnerAccountID, req.Name),
}
h.writeJSON(w, http.StatusOK, resp)
return nil
}