[#1284] writecache: Allow to seal writecache async

Signed-off-by: Dmitrii Stepanov <d.stepanov@yadro.com>

cmd/frostfs-cli/modules/control/writecache.go

@@ -10,6 +10,7 @@ import (
 )
 
 const (
+	asyncFlag       = "async"
 	restoreModeFlag = "restore-mode"
 	shrinkFlag      = "shrink"
 )
@@ -31,12 +32,14 @@ func sealWritecache(cmd *cobra.Command, _ []string) {
 	pk := key.Get(cmd)
 
 	ignoreErrors, _ := cmd.Flags().GetBool(ignoreErrorsFlag)
+	async, _ := cmd.Flags().GetBool(asyncFlag)
 	restoreMode, _ := cmd.Flags().GetBool(restoreModeFlag)
 	shrink, _ := cmd.Flags().GetBool(shrinkFlag)
 
 	req := &control.SealWriteCacheRequest{Body: &control.SealWriteCacheRequest_Body{
 		Shard_ID:     getShardIDList(cmd),
 		IgnoreErrors: ignoreErrors,
+		Async:        async,
 		RestoreMode:  restoreMode,
 		Shrink:       shrink,
 	}}
@@ -77,6 +80,7 @@ func initControlShardsWritecacheCmd() {
 	ff.StringSlice(shardIDFlag, nil, "List of shard IDs in base58 encoding")
 	ff.Bool(shardAllFlag, false, "Process all shards")
 	ff.Bool(ignoreErrorsFlag, true, "Skip invalid/unreadable objects")
+	ff.Bool(asyncFlag, false, "Run operation in background")
 	ff.Bool(restoreModeFlag, false, "Restore writecache's mode after sealing")
 	ff.Bool(shrinkFlag, false, "Shrink writecache's internal storage")
 

internal/logs/logs.go

@@ -539,5 +539,8 @@ const (
 	PolicerCouldNotGetChunk                                              = "could not get EC chunk"
 	PolicerCouldNotGetChunks                                             = "could not get EC chunks"
 	AuditEventLogRecord                                                  = "audit event log record"
+	StartedWritecacheSealAsync                                           = "started writecache seal async"
+	WritecacheSealCompletedAsync                                         = "writecache seal completed successfully"
+	FailedToSealWritecacheAsync                                          = "failed to seal writecache async"
 	WritecacheShrinkSkippedNotEmpty                                      = "writecache shrink skipped: database is not empty"
 )

pkg/local_object_storage/engine/writecache.go

@@ -70,6 +70,7 @@ func (e *StorageEngine) FlushWriteCache(ctx context.Context, p FlushWriteCachePr
 type SealWriteCachePrm struct {
 	ShardIDs     []*shard.ID
 	IgnoreErrors bool
+	Async        bool
 	RestoreMode  bool
 	Shrink       bool
 }
@@ -117,7 +118,7 @@ func (e *StorageEngine) SealWriteCache(ctx context.Context, prm SealWriteCachePr
 				return nil
 			}
 
-			err := sh.SealWriteCache(egCtx, shard.SealWriteCachePrm{IgnoreErrors: prm.IgnoreErrors, RestoreMode: prm.RestoreMode, Shrink: prm.Shrink})
+			err := sh.SealWriteCache(egCtx, shard.SealWriteCachePrm{IgnoreErrors: prm.IgnoreErrors, Async: prm.Async, RestoreMode: prm.RestoreMode, Shrink: prm.Shrink})
 
 			resGuard.Lock()
 			defer resGuard.Unlock()

pkg/local_object_storage/shard/control.go

@@ -178,6 +178,7 @@ func (s *Shard) Init(ctx context.Context) error {
 	if !m.NoMetabase() {
 		s.rb.Start(ctx, s.blobStor, s.metaBase, s.log)
 	}
+	s.writecacheSealCancel.Store(dummyCancel)
 	return nil
 }
 
@@ -350,6 +351,8 @@ func (s *Shard) Close() error {
 	}
 
 	if s.hasWriteCache() {
+		prev := s.writecacheSealCancel.Swap(notInitializedCancel)
+		prev.cancel() // no need to wait: writecache.Seal and writecache.Close lock the same mutex
 		components = append(components, s.writeCache)
 	}
 
@@ -428,6 +431,9 @@ func (s *Shard) lockExclusive() func() {
 		cancelGC := val.(context.CancelFunc)
 		cancelGC()
 	}
+	if c := s.writecacheSealCancel.Load(); c != nil {
+		c.cancel()
+	}
 	s.m.Lock()
 	s.setModeRequested.Store(false)
 	return s.m.Unlock

pkg/local_object_storage/shard/shard.go

@@ -37,8 +37,9 @@ type Shard struct {
 
 	rb *rebuilder
 
-	gcCancel         atomic.Value
-	setModeRequested atomic.Bool
+	gcCancel             atomic.Value
+	setModeRequested     atomic.Bool
+	writecacheSealCancel atomic.Pointer[writecacheSealCanceler]
 }
 
 // Option represents Shard's constructor option.
@@ -190,6 +191,7 @@ func New(opts ...Option) *Shard {
 	}
 
 	s.fillInfo()
+	s.writecacheSealCancel.Store(notInitializedCancel)
 
 	return s
 }

pkg/local_object_storage/shard/writecache.go

@@ -4,12 +4,24 @@ import (
 	"context"
 	"errors"
 
+	"git.frostfs.info/TrueCloudLab/frostfs-node/internal/logs"
 	"git.frostfs.info/TrueCloudLab/frostfs-node/pkg/local_object_storage/writecache"
 	"git.frostfs.info/TrueCloudLab/frostfs-observability/tracing"
 	"go.opentelemetry.io/otel/attribute"
 	"go.opentelemetry.io/otel/trace"
+	"go.uber.org/zap"
 )
 
+var (
+	dummyCancel          = &writecacheSealCanceler{cancel: func() {}}
+	notInitializedCancel = &writecacheSealCanceler{cancel: func() {}}
+	errWriteCacheSealing = errors.New("writecache is already sealing or shard is not initialized")
+)
+
+type writecacheSealCanceler struct {
+	cancel context.CancelFunc
+}
+
 // FlushWriteCachePrm represents parameters of a `FlushWriteCache` operation.
 type FlushWriteCachePrm struct {
 	ignoreErrors bool
@@ -60,6 +72,7 @@ func (s *Shard) FlushWriteCache(ctx context.Context, p FlushWriteCachePrm) error
 
 type SealWriteCachePrm struct {
 	IgnoreErrors bool
+	Async        bool
 	RestoreMode  bool
 	Shrink       bool
 }
@@ -78,15 +91,52 @@ func (s *Shard) SealWriteCache(ctx context.Context, p SealWriteCachePrm) error {
 		return errWriteCacheDisabled
 	}
 
+	if p.Async {
+		ctx = context.WithoutCancel(ctx)
+	}
+	ctx, cancel := context.WithCancel(ctx)
+	canceler := &writecacheSealCanceler{cancel: cancel}
+	if !s.writecacheSealCancel.CompareAndSwap(dummyCancel, canceler) {
+		return errWriteCacheSealing
+	}
 	s.m.RLock()
-	defer s.m.RUnlock()
+	cleanup := func() {
+		s.m.RUnlock()
+		s.writecacheSealCancel.Store(dummyCancel)
+	}
 
 	if s.info.Mode.ReadOnly() {
+		cleanup()
 		return ErrReadOnlyMode
 	}
 	if s.info.Mode.NoMetabase() {
+		cleanup()
 		return ErrDegradedMode
 	}
 
-	return s.writeCache.Seal(ctx, writecache.SealPrm{IgnoreErrors: p.IgnoreErrors, RestoreMode: p.RestoreMode, Shrink: p.Shrink})
+	if !p.Async {
+		defer cleanup()
+	}
+	prm := writecache.SealPrm{IgnoreErrors: p.IgnoreErrors, RestoreMode: p.RestoreMode, Shrink: p.Shrink}
+	if p.Async {
+		started := make(chan struct{})
+		go func() {
+			close(started)
+			defer cleanup()
+
+			s.log.Info(logs.StartedWritecacheSealAsync)
+			if err := s.writeCache.Seal(ctx, prm); err != nil {
+				s.log.Warn(logs.FailedToSealWritecacheAsync, zap.Error(err))
+				return
+			}
+			s.log.Info(logs.WritecacheSealCompletedAsync)
+		}()
+		select {
+		case <-ctx.Done():
+			return ctx.Err()
+		case <-started:
+			return nil
+		}
+	}
+	return s.writeCache.Seal(ctx, prm)
 }

pkg/services/control/server/seal_writecache.go

@@ -19,6 +19,7 @@ func (s *Server) SealWriteCache(ctx context.Context, req *control.SealWriteCache
 	prm := engine.SealWriteCachePrm{
 		ShardIDs:     s.getShardIDList(req.GetBody().GetShard_ID()),
 		IgnoreErrors: req.GetBody().GetIgnoreErrors(),
+		Async:        req.GetBody().GetAsync(),
 		RestoreMode:  req.GetBody().GetRestoreMode(),
 		Shrink:       req.GetBody().GetShrink(),
 	}

pkg/services/control/service.proto

@@ -656,6 +656,9 @@ message SealWriteCacheRequest {
     // Flag indicating whether object read errors should be ignored.
     bool ignore_errors = 2;
 
+    // Flag indicating whether writecache will be sealed async.
+    bool async = 3;
+
     // If true, then writecache will be sealed, but mode will be restored to the current one.
     bool restore_mode = 4;