Files
seaweedfs/weed/mq/offset/consumer_group_storage.go
T
Chris LuandGitHub 35e3fe89bc feat(s3/lifecycle): filer-backed cursor Persister + drop BlockerStore (#9358)
* feat(s3/lifecycle): filer-backed cursor Persister

FilerPersister persists per-shard cursor maps as JSON to
/etc/s3/lifecycle/cursors/shard-NN.json via filer.SaveInsideFiler.
One file per shard keeps Save atomic — the filer writes the entry
in a single mutation, so a crash mid-write doesn't leak partial
state. Pipeline.Run loads on start; the periodic checkpoint and
graceful-shutdown save go through this implementation.

A small FilerStore interface wraps the SeaweedFilerClient surface
the persister needs, so tests inject an in-memory fake instead of
mocking the full gRPC client.

* refactor(s3/lifecycle): drop BlockerStore — durable cursor IS the block

A frozen cursor doesn't advance, so the durable cursor (FilerPersister)
encodes the blocked state on its own. On worker restart the reader
re-encounters the poison event at MinTsNs, the dispatcher walks the
same retry budget to BLOCKED, and the cursor freezes at the same
EventTs. Other in-flight events between freeze tsNs and prior cursor
positions self-resolve via NOOP_RESOLVED (STALE_IDENTITY) since the
underlying objects were already deleted on the prior pass.

Removed:
  - BlockerStore interface + InMemoryBlockerStore + BlockerRecord
  - Dispatcher.Blockers + Dispatcher.ReplayBlockers
  - the BlockerStore.Put call in handleBlocked
  - Pipeline.Blockers field + the ReplayBlockers call on startup

Added a TestDispatchRestartReFreezesNaturally that pins the
self-recovery property: a fresh Dispatcher with a fresh Cursor, fed
the same poison event, reaches the same frozen state at the same
EventTs without any durable blocker store.

Operator visibility: a cursor whose MinTsNs hasn't advanced is the
signal — surfaced via the durable cursor file.

* refactor(filer): SaveInsideFiler accepts ctx

ReadInsideFiler already takes ctx; SaveInsideFiler used context.Background()
internally and silently dropped the caller's ctx. Symmetric API now;
cancellation/deadlines propagate through LookupEntry / CreateEntry /
UpdateEntry. Mechanical update of all callers — most pass
context.Background() since the existing call sites have no ctx in scope.

* fix(s3/lifecycle): deterministic order in cursor save

Iterating Go maps yields random order, so json.Encode produced a different
byte sequence on each save even when the state hadn't changed. Sort
entries by (Bucket, ActionKind, RuleHash) before encoding so the on-disk
file diffs cleanly. New test pins byte-identical output across two saves
of the same map.

* fix(s3/lifecycle): log reason when freezing cursor in handleBlocked

handleBlocked dropped the reason via _ = reason with a comment claiming
the caller logged it; none of the three callers do. A frozen cursor is
the only surface where the operator finds out something stuck, so the
reason has to land somewhere. glog.Warningf with shard, key, eventTs,
and the original reason — same shape the rest of the package uses.
2026-05-07 17:45:04 -07:00

182 lines
7.4 KiB
Go

package offset
import (
"context"
"encoding/json"
"fmt"
"io"
"time"
"github.com/seaweedfs/seaweedfs/weed/filer"
"github.com/seaweedfs/seaweedfs/weed/filer_client"
"github.com/seaweedfs/seaweedfs/weed/mq/topic"
"github.com/seaweedfs/seaweedfs/weed/pb/filer_pb"
"github.com/seaweedfs/seaweedfs/weed/pb/schema_pb"
)
// ConsumerGroupPosition represents a consumer's position in a partition
// This can be either a timestamp or an offset
type ConsumerGroupPosition struct {
Type string `json:"type"` // "offset" or "timestamp"
Value int64 `json:"value"` // The actual offset or timestamp value
OffsetType string `json:"offset_type"` // Optional: OffsetType enum name (e.g., "EXACT_OFFSET")
CommittedAt int64 `json:"committed_at"` // Unix timestamp in milliseconds when committed
Metadata string `json:"metadata"` // Optional: application-specific metadata
}
// ConsumerGroupOffsetStorage handles consumer group offset persistence
// Each consumer group gets its own offset file in a dedicated consumers/ subfolder:
// Path: /topics/{namespace}/{topic}/{version}/{partition}/consumers/{consumer_group}.offset
type ConsumerGroupOffsetStorage interface {
// SaveConsumerGroupOffset saves the committed offset for a consumer group
SaveConsumerGroupOffset(t topic.Topic, p topic.Partition, consumerGroup string, offset int64) error
// SaveConsumerGroupPosition saves the committed position (offset or timestamp) for a consumer group
SaveConsumerGroupPosition(t topic.Topic, p topic.Partition, consumerGroup string, position *ConsumerGroupPosition) error
// LoadConsumerGroupOffset loads the committed offset for a consumer group (backward compatible)
LoadConsumerGroupOffset(t topic.Topic, p topic.Partition, consumerGroup string) (int64, error)
// LoadConsumerGroupPosition loads the committed position for a consumer group
LoadConsumerGroupPosition(t topic.Topic, p topic.Partition, consumerGroup string) (*ConsumerGroupPosition, error)
// ListConsumerGroups returns all consumer groups for a topic partition
ListConsumerGroups(t topic.Topic, p topic.Partition) ([]string, error)
// DeleteConsumerGroupOffset removes the offset file for a consumer group
DeleteConsumerGroupOffset(t topic.Topic, p topic.Partition, consumerGroup string) error
}
// FilerConsumerGroupOffsetStorage implements ConsumerGroupOffsetStorage using SeaweedFS filer
type FilerConsumerGroupOffsetStorage struct {
filerClientAccessor *filer_client.FilerClientAccessor
}
// NewFilerConsumerGroupOffsetStorageWithAccessor creates storage using a shared filer client accessor
func NewFilerConsumerGroupOffsetStorageWithAccessor(filerClientAccessor *filer_client.FilerClientAccessor) *FilerConsumerGroupOffsetStorage {
return &FilerConsumerGroupOffsetStorage{
filerClientAccessor: filerClientAccessor,
}
}
// SaveConsumerGroupOffset saves the committed offset for a consumer group
// Stores as: /topics/{namespace}/{topic}/{version}/{partition}/consumers/{consumer_group}.offset
// This is a convenience method that wraps SaveConsumerGroupPosition
func (f *FilerConsumerGroupOffsetStorage) SaveConsumerGroupOffset(t topic.Topic, p topic.Partition, consumerGroup string, offset int64) error {
position := &ConsumerGroupPosition{
Type: "offset",
Value: offset,
OffsetType: schema_pb.OffsetType_EXACT_OFFSET.String(),
CommittedAt: time.Now().UnixMilli(),
}
return f.SaveConsumerGroupPosition(t, p, consumerGroup, position)
}
// SaveConsumerGroupPosition saves the committed position (offset or timestamp) for a consumer group
// Stores as JSON: /topics/{namespace}/{topic}/{version}/{partition}/consumers/{consumer_group}.offset
func (f *FilerConsumerGroupOffsetStorage) SaveConsumerGroupPosition(t topic.Topic, p topic.Partition, consumerGroup string, position *ConsumerGroupPosition) error {
partitionDir := topic.PartitionDir(t, p)
consumersDir := fmt.Sprintf("%s/consumers", partitionDir)
offsetFileName := fmt.Sprintf("%s.offset", consumerGroup)
// Marshal position to JSON
jsonBytes, err := json.Marshal(position)
if err != nil {
return fmt.Errorf("failed to marshal position to JSON: %w", err)
}
return f.filerClientAccessor.WithFilerClient(false, func(client filer_pb.SeaweedFilerClient) error {
return filer.SaveInsideFiler(context.Background(), client, consumersDir, offsetFileName, jsonBytes)
})
}
// LoadConsumerGroupOffset loads the committed offset for a consumer group
// This method provides backward compatibility and returns just the offset value
func (f *FilerConsumerGroupOffsetStorage) LoadConsumerGroupOffset(t topic.Topic, p topic.Partition, consumerGroup string) (int64, error) {
position, err := f.LoadConsumerGroupPosition(t, p, consumerGroup)
if err != nil {
return -1, err
}
return position.Value, nil
}
// LoadConsumerGroupPosition loads the committed position for a consumer group
func (f *FilerConsumerGroupOffsetStorage) LoadConsumerGroupPosition(t topic.Topic, p topic.Partition, consumerGroup string) (*ConsumerGroupPosition, error) {
partitionDir := topic.PartitionDir(t, p)
consumersDir := fmt.Sprintf("%s/consumers", partitionDir)
offsetFileName := fmt.Sprintf("%s.offset", consumerGroup)
var position *ConsumerGroupPosition
err := f.filerClientAccessor.WithFilerClient(false, func(client filer_pb.SeaweedFilerClient) error {
data, err := filer.ReadInsideFiler(context.Background(), client, consumersDir, offsetFileName)
if err != nil {
return err
}
// Parse JSON format
position = &ConsumerGroupPosition{}
if err := json.Unmarshal(data, position); err != nil {
return fmt.Errorf("invalid consumer group offset file format: %w", err)
}
return nil
})
if err != nil {
return nil, err
}
return position, nil
}
// ListConsumerGroups returns all consumer groups for a topic partition
func (f *FilerConsumerGroupOffsetStorage) ListConsumerGroups(t topic.Topic, p topic.Partition) ([]string, error) {
partitionDir := topic.PartitionDir(t, p)
consumersDir := fmt.Sprintf("%s/consumers", partitionDir)
var consumerGroups []string
err := f.filerClientAccessor.WithFilerClient(false, func(client filer_pb.SeaweedFilerClient) error {
// Use ListEntries to get directory contents
stream, err := client.ListEntries(context.Background(), &filer_pb.ListEntriesRequest{
Directory: consumersDir,
})
if err != nil {
return err
}
for {
resp, err := stream.Recv()
if err != nil {
if err == io.EOF {
break
}
return err
}
entry := resp.Entry
if entry != nil && !entry.IsDirectory && entry.Name != "" {
// Check if this is a consumer group offset file (ends with .offset)
if len(entry.Name) > 7 && entry.Name[len(entry.Name)-7:] == ".offset" {
// Extract consumer group name (remove .offset suffix)
consumerGroup := entry.Name[:len(entry.Name)-7]
consumerGroups = append(consumerGroups, consumerGroup)
}
}
}
return nil
})
return consumerGroups, err
}
// DeleteConsumerGroupOffset removes the offset file for a consumer group
func (f *FilerConsumerGroupOffsetStorage) DeleteConsumerGroupOffset(t topic.Topic, p topic.Partition, consumerGroup string) error {
partitionDir := topic.PartitionDir(t, p)
consumersDir := fmt.Sprintf("%s/consumers", partitionDir)
offsetFileName := fmt.Sprintf("%s.offset", consumerGroup)
return f.filerClientAccessor.WithFilerClient(false, func(client filer_pb.SeaweedFilerClient) error {
return filer_pb.DoRemove(context.Background(), client, consumersDir, offsetFileName, false, false, false, false, nil)
})
}