milvus/internal/storage/gcp_native_object_storage.go
wei liu 975c91df16
feat: Add comprehensive snapshot functionality for collections (#44361)
issue: #44358

Implement complete snapshot management system including creation,
deletion, listing, description, and restoration capabilities across all
system components.

Key features:
- Create snapshots for entire collections
- Drop snapshots by name with proper cleanup
- List snapshots with collection filtering
- Describe snapshot details and metadata

Components added/modified:
- Client SDK with full snapshot API support and options
- DataCoord snapshot service with metadata management
- Proxy layer with task-based snapshot operations
- Protocol buffer definitions for snapshot RPCs
- Comprehensive unit tests with mockey framework
- Integration tests for end-to-end validation

Technical implementation:
- Snapshot metadata storage in etcd with proper indexing
- File-based snapshot data persistence in object storage
- Garbage collection integration for snapshot cleanup
- Error handling and validation across all operations
- Thread-safe operations with proper locking mechanisms

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
- Core invariant/assumption: snapshots are immutable point‑in‑time
captures identified by (collection, snapshot name/ID); etcd snapshot
metadata is authoritative for lifecycle (PENDING → COMMITTED → DELETING)
and per‑segment manifests live in object storage (Avro / StorageV2). GC
and restore logic must see snapshotRefIndex loaded
(snapshotMeta.IsRefIndexLoaded) before reclaiming or relying on
segment/index files.

- New capability added: full end‑to‑end snapshot subsystem — client SDK
APIs (Create/Drop/List/Describe/Restore + restore job queries),
DataCoord SnapshotWriter/Reader (Avro + StorageV2 manifests),
snapshotMeta in meta, SnapshotManager orchestration
(create/drop/describe/list/restore), copy‑segment restore
tasks/inspector/checker, proxy & RPC surface, GC integration, and
docs/tests — enabling point‑in‑time collection snapshots persisted to
object storage and restorations orchestrated across components.

- Logic removed/simplified and why: duplicated recursive
compaction/delta‑log traversal and ad‑hoc lookup code were consolidated
behind two focused APIs/owners (Handler.GetDeltaLogFromCompactTo for
delta traversal and SnapshotManager/SnapshotReader for snapshot I/O).
MixCoord/coordinator broker paths were converted to thin RPC proxies.
This eliminates multiple implementations of the same traversal/lookup,
reducing divergence and simplifying responsibility boundaries.

- Why this does NOT introduce data loss or regressions: snapshot
create/drop use explicit two‑phase semantics (PENDING → COMMIT/DELETING)
with SnapshotWriter writing manifests and metadata before commit; GC
uses snapshotRefIndex guards and
IsRefIndexLoaded/GetSnapshotBySegment/GetSnapshotByIndex checks to avoid
removing referenced files; restore flow pre‑allocates job IDs, validates
resources (partitions/indexes), performs rollback on failure
(rollbackRestoreSnapshot), and converts/updates segment/index metadata
only after successful copy tasks. Extensive unit and integration tests
exercise pending/deleting/GC/restore/error paths to ensure idempotence
and protection against premature deletion.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->

---------

Signed-off-by: Wei Liu <wei.liu@zilliz.com>
2026-01-06 10:15:24 +08:00

260 lines
6.6 KiB
Go

// Licensed to the LF AI & Data foundation under one
// or more contributor license agreements. See the NOTICE file
// distributed with this work for additional information
// regarding copyright ownership. The ASF licenses this file
// to you under the Apache License, Version 2.0 (the
// "License"); you may not use this file except in compliance
// with the License. You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package storage
import (
"context"
"io"
"cloud.google.com/go/storage"
"google.golang.org/api/googleapi"
"google.golang.org/api/iterator"
"github.com/milvus-io/milvus/pkg/v2/objectstorage"
"github.com/milvus-io/milvus/pkg/v2/util/merr"
)
type GcpNativeObjectStorage struct {
client *storage.Client
}
func newGcpNativeObjectStorageWithConfig(ctx context.Context, c *objectstorage.Config) (*GcpNativeObjectStorage, error) {
client, err := objectstorage.NewGcpObjectStorageClient(ctx, c)
if err != nil {
return nil, err
}
return &GcpNativeObjectStorage{client: client}, nil
}
func (gcs *GcpNativeObjectStorage) GetObject(ctx context.Context, bucketName, objectName string,
offset int64, size int64,
) (FileReader, error) {
bucket := gcs.client.Bucket(bucketName)
_, err := bucket.Attrs(ctx)
if err != nil {
return nil, checkObjectStorageError(objectName, err)
}
obj := bucket.Object(objectName)
_, err = obj.Attrs(ctx)
if err != nil {
return nil, checkObjectStorageError(objectName, err)
}
var reader *storage.Reader
if offset == 0 && size == 0 {
reader, err = obj.NewReader(ctx)
} else {
reader, err = obj.NewRangeReader(ctx, offset, size)
}
if err != nil {
return nil, checkObjectStorageError(objectName, err)
}
return &GcsReader{
reader: reader,
obj: obj,
objectSize: reader.Attrs.Size,
}, nil
}
func (gcs *GcpNativeObjectStorage) PutObject(ctx context.Context, bucketName, objectName string,
reader io.Reader, objectSize int64,
) error {
obj := gcs.client.Bucket(bucketName).Object(objectName)
writer := obj.NewWriter(ctx)
_, err := io.Copy(writer, reader)
if err != nil {
return checkObjectStorageError(objectName, err)
}
err = writer.Close()
if err != nil {
return checkObjectStorageError(objectName, err)
}
return nil
}
func (gcs *GcpNativeObjectStorage) StatObject(ctx context.Context, bucketName,
objectName string,
) (int64, error) {
obj := gcs.client.Bucket(bucketName).Object(objectName)
attrs, err := obj.Attrs(ctx)
if err != nil {
return 0, checkObjectStorageError(objectName, err)
}
return attrs.Size, nil
}
func (gcs *GcpNativeObjectStorage) WalkWithObjects(ctx context.Context,
bucketName string, prefix string, recursive bool, walkFunc ChunkObjectWalkFunc,
) error {
query := &storage.Query{
Prefix: prefix,
}
if !recursive {
query.Delimiter = "/"
}
it := gcs.client.Bucket(bucketName).Objects(ctx, query)
for {
objAttrs, err := it.Next()
if err == iterator.Done {
break
}
if err != nil {
return checkObjectStorageError(prefix, err)
}
if objAttrs.Prefix != "" {
if !walkFunc(&ChunkObjectInfo{FilePath: objAttrs.Prefix, ModifyTime: objAttrs.Updated}) {
return nil
}
} else if objAttrs.Name != "" {
if !walkFunc(&ChunkObjectInfo{FilePath: objAttrs.Name, ModifyTime: objAttrs.Updated}) {
return nil
}
}
}
return nil
}
func (gcs *GcpNativeObjectStorage) RemoveObject(ctx context.Context, bucketName, prefix string) error {
bucket := gcs.client.Bucket(bucketName)
query := &storage.Query{Prefix: prefix}
it := bucket.Objects(ctx, query)
for {
objAttrs, err := it.Next()
if err == iterator.Done {
break
}
if err != nil {
return checkObjectStorageError(prefix, err)
}
obj := bucket.Object(objAttrs.Name)
if err := obj.Delete(ctx); err != nil {
return checkObjectStorageError(objAttrs.Name, err)
}
}
return nil
}
func (gcs *GcpNativeObjectStorage) CopyObject(ctx context.Context, bucketName, srcObjectName, dstObjectName string) error {
bucket := gcs.client.Bucket(bucketName)
srcObj := bucket.Object(srcObjectName)
dstObj := bucket.Object(dstObjectName)
// Use CopierFrom to copy object
_, err := dstObj.CopierFrom(srcObj).Run(ctx)
return checkObjectStorageError(dstObjectName, err)
}
func (gcs *GcpNativeObjectStorage) DeleteBucket(ctx context.Context, bucketName string) error {
bucket := gcs.client.Bucket(bucketName)
err := gcs.RemoveObject(ctx, bucketName, "")
if err != nil {
return err
}
err = bucket.Delete(ctx)
if err != nil {
return err
}
return nil
}
type GcsReader struct {
reader *storage.Reader
obj *storage.ObjectHandle
position int64
objectSize int64
}
func (gcsReader *GcsReader) Read(p []byte) (n int, err error) {
n, err = gcsReader.reader.Read(p)
if err != nil {
return n, err
}
gcsReader.position = gcsReader.position + int64(n)
return n, nil
}
func (gcsReader *GcsReader) Close() error {
return gcsReader.reader.Close()
}
func (gcsReader *GcsReader) ReadAt(p []byte, off int64) (n int, err error) {
reader, err := gcsReader.obj.NewRangeReader(context.Background(), off, int64(len(p)))
if err != nil {
return 0, err
}
defer reader.Close()
return io.ReadFull(reader, p)
}
func (gcsReader *GcsReader) Seek(offset int64, whence int) (int64, error) {
var newOffset int64
switch whence {
case io.SeekStart:
newOffset = offset
case io.SeekCurrent:
newOffset = gcsReader.position + offset
case io.SeekEnd:
objectAttrs, err := gcsReader.obj.Attrs(context.Background())
if err != nil {
return 0, err
}
newOffset = objectAttrs.Size + offset
default:
return 0, merr.WrapErrIoFailedReason("invalid whence")
}
if newOffset < 0 {
return 0, merr.WrapErrIoFailedReason("negative offset")
}
// Reset the underlying reader to the new offset
newReader, err := gcsReader.obj.NewRangeReader(context.Background(), newOffset, -1)
if err != nil {
if gErr, ok := err.(*googleapi.Error); ok {
if gErr.Code == 416 {
newReader, _ = gcsReader.obj.NewRangeReader(context.Background(), 0, 0)
}
} else {
return 0, err
}
}
if gcsReader.reader != nil {
if err := gcsReader.reader.Close(); err != nil {
return 0, err
}
}
// Update the reader and the current position
gcsReader.reader = newReader
gcsReader.position = newOffset
return newOffset, nil
}
func (gcsReader *GcsReader) Size() (int64, error) {
return gcsReader.objectSize, nil
}