mirror of
https://github.com/seaweedfs/seaweedfs.git
synced 2026-09-10 08:30:47 +02:00
* s3: give a versioned metadata-only copy its own chunks A self-copy that only rewrites metadata clones the source entry, chunk fids and all, and writes the clone back. With no versioning that is exactly right: the clone replaces the entry it came from, so one entry owns the needles the whole time. Under versioning the clone lands in a new .versions/ file and the source stays live, and nothing refcounts a plain shared chunk list -- deleting either version (a NoncurrentVersionExpiration rule, say) frees needles the other still points at, and the next vacuum makes that permanent. rclone hits this on every upload, since it stamps mtime with exactly this copy. Take the metadata-only path only where the write replaces the entry it read: the bare key of a bucket without versioning. Versioned, suspended, and versionId-pinned copies fall through to the regular copy path, which gives the destination its own chunks. * s3: reencrypt a versioned SSE-KMS key rotation instead of reusing the chunks A same-object copy that changes the KMS key id hands the source chunks straight back, on the assumption that the copy overwrites the entry they came from. A versioned bucket writes a new version beside the source instead, so the two end up sharing needles that nothing refcounts, and deleting either one frees the other's data. Reuse the chunks only when the destination really is the source entry; otherwise fall through to the reencrypt path, which also gives the new version the key it asked for rather than leaving it on the old one. * s3: make one predicate decide whether a copy replaces its source The metadata-only branch and the key-rotation strategy both answer the same question -- does this copy write back to the entry it read -- so let them share one predicate instead of pairing a same-destination check with it separately at each site. * test(s3): fail the copy regression tests when the vacuum does not run The helper swallowed a failed or non-200 request to the master, so a vacuum that never ran turned both chunk-ownership assertions into no-ops: the tombstoned needles were still readable and the surviving version looked fine either way. Require the endpoint, the request, and a 200. * ci(s3): run every versioning test in the regression gate The gate named the tests it wanted, so a new regression test sat there uncovered until someone remembered this file -- it fooled me into thinking two tests added in this PR never ran anywhere, when the comprehensive job had them all along. Invert it: run everything, and name a test only to keep it out. The delete job beside this one already works that way, and the suite costs about two minutes. Only the pagination stress tests are excluded; they build 1500+ versions, skip themselves without ENABLE_STRESS_TESTS, and have their own make target. Go's regexp has no negation, so the pattern is still assembled from a listing, the way the volume-server integration workflow does it. Note the trailing $$: make eats a lone trailing $ and takes the anchor with it.
188 lines
7.7 KiB
Go
188 lines
7.7 KiB
Go
package s3api
|
|
|
|
import (
|
|
"errors"
|
|
"fmt"
|
|
"net/http"
|
|
|
|
"github.com/seaweedfs/seaweedfs/weed/glog"
|
|
"github.com/seaweedfs/seaweedfs/weed/pb/filer_pb"
|
|
"github.com/seaweedfs/seaweedfs/weed/s3api/s3err"
|
|
weed_server "github.com/seaweedfs/seaweedfs/weed/server"
|
|
)
|
|
|
|
// executeUnifiedCopyStrategy executes the appropriate copy strategy based on encryption state
|
|
// Returns chunks and destination metadata that should be applied to the destination entry.
|
|
// replacesSource says the destination entry is the source entry, which is what lets the
|
|
// key-rotation strategy hand back the source chunks instead of copying them.
|
|
func (s3a *S3ApiServer) executeUnifiedCopyStrategy(entry *filer_pb.Entry, r *http.Request, srcBucket, dstBucket, srcObject, dstObject string, replacesSource bool) ([]*filer_pb.FileChunk, map[string][]byte, error) {
|
|
// Per-chunk copy must see data chunks: a manifest chunk copied raw becomes
|
|
// object data. Resolved manifests stay with the source.
|
|
if _, err := s3a.flattenManifestChunks(r.Context(), entry); err != nil {
|
|
return nil, nil, fmt.Errorf("resolve source manifest chunks: %w", err)
|
|
}
|
|
|
|
// Detect encryption state (using entry-aware detection for multipart objects)
|
|
srcPath := fmt.Sprintf("%s/%s", s3a.bucketDir(srcBucket), srcObject)
|
|
dstPath := fmt.Sprintf("%s/%s", s3a.bucketDir(dstBucket), dstObject)
|
|
state := DetectEncryptionStateWithEntry(entry, r, srcPath, dstPath)
|
|
|
|
// Debug logging for encryption state
|
|
|
|
// Apply bucket default encryption if no explicit encryption specified
|
|
s3a.applyCopyBucketDefaultEncryption(state, dstBucket)
|
|
|
|
// Determine copy strategy
|
|
strategy, err := DetermineUnifiedCopyStrategy(state, entry.Extended, r)
|
|
if err != nil {
|
|
return nil, nil, err
|
|
}
|
|
|
|
glog.V(2).Infof("Unified copy strategy for %s → %s: %v", srcPath, dstPath, strategy)
|
|
|
|
// Calculate optimized sizes for the strategy
|
|
sizeCalc := CalculateOptimizedSizes(entry, r, strategy)
|
|
glog.V(2).Infof("Size calculation: src=%d, target=%d, actual=%d, overhead=%d, preallocate=%v",
|
|
sizeCalc.SourceSize, sizeCalc.TargetSize, sizeCalc.ActualContentSize,
|
|
sizeCalc.EncryptionOverhead, sizeCalc.CanPreallocate)
|
|
|
|
// Execute strategy
|
|
switch strategy {
|
|
case CopyStrategyDirect:
|
|
chunks, err := s3a.copyChunks(entry, dstPath)
|
|
return chunks, nil, err
|
|
|
|
case CopyStrategyKeyRotation:
|
|
return s3a.executeKeyRotation(entry, r, state, dstBucket, dstPath, replacesSource)
|
|
|
|
case CopyStrategyEncrypt:
|
|
return s3a.executeEncryptCopy(entry, r, state, dstBucket, dstPath)
|
|
|
|
case CopyStrategyDecrypt:
|
|
return s3a.executeDecryptCopy(entry, r, state, dstPath)
|
|
|
|
case CopyStrategyReencrypt:
|
|
return s3a.executeReencryptCopy(entry, r, state, dstBucket, dstPath)
|
|
|
|
default:
|
|
return nil, nil, fmt.Errorf("unknown unified copy strategy: %v", strategy)
|
|
}
|
|
}
|
|
|
|
// mapCopyErrorToS3Error maps various copy errors to appropriate S3 error codes
|
|
func (s3a *S3ApiServer) mapCopyErrorToS3Error(err error) s3err.ErrorCode {
|
|
if err == nil {
|
|
return s3err.ErrNone
|
|
}
|
|
|
|
// Check for read-only errors (quota enforcement)
|
|
// Uses errors.Is() to properly detect wrapped errors
|
|
if errors.Is(err, weed_server.ErrReadOnly) {
|
|
// Bucket is read-only due to quota enforcement or other configuration
|
|
// Return 403 Forbidden per S3 semantics (similar to MinIO's quota enforcement)
|
|
return s3err.ErrAccessDenied
|
|
}
|
|
|
|
// Check for KMS errors first
|
|
if kmsErr := MapKMSErrorToS3Error(err); kmsErr != s3err.ErrInvalidRequest {
|
|
return kmsErr
|
|
}
|
|
|
|
// Check for SSE-C errors
|
|
if ssecErr := MapSSECErrorToS3Error(err); ssecErr != s3err.ErrInvalidRequest {
|
|
return ssecErr
|
|
}
|
|
|
|
// Default to internal error for unknown errors
|
|
return s3err.ErrInternalError
|
|
}
|
|
|
|
// executeKeyRotation handles key rotation for same-object copies
|
|
func (s3a *S3ApiServer) executeKeyRotation(entry *filer_pb.Entry, r *http.Request, state *EncryptionState, dstBucket, dstPath string, replacesSource bool) ([]*filer_pb.FileChunk, map[string][]byte, error) {
|
|
// For key rotation, we only need to update metadata, not re-copy chunks
|
|
// This is a significant optimization for same-object key changes
|
|
|
|
if state.SrcSSEC && state.DstSSEC {
|
|
// SSE-C key rotation - need to handle new key/IV, use reencrypt logic
|
|
return s3a.executeReencryptCopy(entry, r, state, dstBucket, dstPath)
|
|
}
|
|
|
|
// Handing back the source chunks leaves the destination sharing needles nothing
|
|
// refcounts, so it is only safe when the destination overwrites the source entry.
|
|
// A versioned rotation writes a new version beside the source and has to reencrypt.
|
|
if state.SrcSSEKMS && state.DstSSEKMS && replacesSource {
|
|
// SSE-KMS key rotation - return existing chunks, metadata will be updated by caller
|
|
return entry.GetChunks(), nil, nil
|
|
}
|
|
|
|
// Fallback to reencrypt if we can't do metadata-only rotation
|
|
return s3a.executeReencryptCopy(entry, r, state, dstBucket, dstPath)
|
|
}
|
|
|
|
// executeEncryptCopy handles plain → encrypted copies
|
|
func (s3a *S3ApiServer) executeEncryptCopy(entry *filer_pb.Entry, r *http.Request, state *EncryptionState, dstBucket, dstPath string) ([]*filer_pb.FileChunk, map[string][]byte, error) {
|
|
if state.DstSSEC {
|
|
// Use existing SSE-C copy logic
|
|
return s3a.copyChunksWithSSEC(entry, r, dstPath)
|
|
}
|
|
|
|
if state.DstSSEKMS {
|
|
// Use existing SSE-KMS copy logic - metadata is now generated internally
|
|
chunks, dstMetadata, err := s3a.copyChunksWithSSEKMS(entry, r, dstBucket, dstPath)
|
|
return chunks, dstMetadata, err
|
|
}
|
|
|
|
if state.DstSSES3 {
|
|
// Use chunk-by-chunk copy for SSE-S3 encryption (consistent with SSE-C and SSE-KMS)
|
|
glog.V(2).Infof("Plain→SSE-S3 copy: using unified multipart encrypt copy")
|
|
return s3a.copyMultipartCrossEncryption(entry, r, state, dstBucket, dstPath)
|
|
}
|
|
|
|
return nil, nil, fmt.Errorf("unknown target encryption type")
|
|
}
|
|
|
|
// executeDecryptCopy handles encrypted → plain copies
|
|
func (s3a *S3ApiServer) executeDecryptCopy(entry *filer_pb.Entry, r *http.Request, state *EncryptionState, dstPath string) ([]*filer_pb.FileChunk, map[string][]byte, error) {
|
|
// Use unified multipart-aware decrypt copy for all encryption types (consistent chunk-by-chunk)
|
|
if state.SrcSSEC || state.SrcSSEKMS || state.SrcSSES3 {
|
|
glog.V(2).Infof("Encrypted→Plain copy: using unified multipart decrypt copy")
|
|
return s3a.copyMultipartCrossEncryption(entry, r, state, "", dstPath)
|
|
}
|
|
|
|
return nil, nil, fmt.Errorf("unknown source encryption type")
|
|
}
|
|
|
|
// executeReencryptCopy handles encrypted → encrypted copies with different keys/methods
|
|
func (s3a *S3ApiServer) executeReencryptCopy(entry *filer_pb.Entry, r *http.Request, state *EncryptionState, dstBucket, dstPath string) ([]*filer_pb.FileChunk, map[string][]byte, error) {
|
|
// Use chunk-by-chunk approach for all cross-encryption scenarios (consistent behavior)
|
|
if state.SrcSSEC && state.DstSSEC {
|
|
return s3a.copyChunksWithSSEC(entry, r, dstPath)
|
|
}
|
|
|
|
if state.SrcSSEKMS && state.DstSSEKMS {
|
|
// Use existing SSE-KMS copy logic - metadata is now generated internally
|
|
chunks, dstMetadata, err := s3a.copyChunksWithSSEKMS(entry, r, dstBucket, dstPath)
|
|
return chunks, dstMetadata, err
|
|
}
|
|
|
|
// All other cross-encryption scenarios use unified multipart copy
|
|
// This includes: SSE-C↔SSE-KMS, SSE-C↔SSE-S3, SSE-KMS↔SSE-S3, SSE-S3↔SSE-S3
|
|
glog.V(2).Infof("Cross-encryption copy: using unified multipart copy")
|
|
return s3a.copyMultipartCrossEncryption(entry, r, state, dstBucket, dstPath)
|
|
}
|
|
|
|
// applyCopyBucketDefaultEncryption applies the destination bucket's default encryption settings if no explicit encryption is specified
|
|
func (s3a *S3ApiServer) applyCopyBucketDefaultEncryption(state *EncryptionState, dstBucket string) {
|
|
if !state.IsTargetEncrypted() {
|
|
bucketMetadata, err := s3a.getBucketMetadata(dstBucket)
|
|
if err == nil && bucketMetadata != nil && bucketMetadata.Encryption != nil {
|
|
switch bucketMetadata.Encryption.SseAlgorithm {
|
|
case "aws:kms":
|
|
state.DstSSEKMS = true
|
|
case "AES256":
|
|
state.DstSSES3 = true
|
|
}
|
|
}
|
|
}
|
|
}
|