fix(BatchPublishing): Make topic.AddToBatch threadsafe (#622)

topic.Publish is already thread safe. topic.AddToBatch should strive to
follow similar semantics.

Looking at how this would integrate with Prysm, they use separate
goroutines per message they'd like to batch.
This commit is contained in:
Marco Munizaga 2025-06-25 12:38:21 -07:00 committed by GitHub
parent 3f89e4331c
commit fedbccc0c6
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
4 changed files with 86 additions and 53 deletions

View File

@ -3682,6 +3682,9 @@ func BenchmarkRoundRobinMessageIDScheduler(b *testing.B) {
}
func TestMessageBatchPublish(t *testing.T) {
concurrentAdds := []bool{false, true}
for _, concurrentAdd := range concurrentAdds {
t.Run(fmt.Sprintf("WithConcurrentAdd=%v", concurrentAdd), func(t *testing.T) {
ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
defer cancel()
hosts := getDefaultHosts(t, 20)
@ -3718,13 +3721,27 @@ func TestMessageBatchPublish(t *testing.T) {
time.Sleep(time.Second * 2)
var batch MessageBatch
var wg sync.WaitGroup
for i := 0; i < numMessages; i++ {
msg := []byte(fmt.Sprintf("%d it's not a floooooood %d", i, i))
if concurrentAdd {
wg.Add(1)
go func() {
defer wg.Done()
err := topics[0].AddToBatch(ctx, &batch, msg)
if err != nil {
t.Log(err)
t.Fail()
}
}()
} else {
err := topics[0].AddToBatch(ctx, &batch, msg)
if err != nil {
t.Fatal(err)
}
}
}
wg.Wait()
err := psubs[0].PublishBatch(&batch)
if err != nil {
t.Fatal(err)
@ -3743,6 +3760,8 @@ func TestMessageBatchPublish(t *testing.T) {
}
}
}
})
}
}
func TestPublishDuplicateMessage(t *testing.T) {

View File

@ -2,6 +2,7 @@ package pubsub
import (
"iter"
"sync"
"github.com/libp2p/go-libp2p/core/peer"
)
@ -10,9 +11,24 @@ import (
// once. This allows the Scheduler to define an order for outgoing RPCs.
// This helps bandwidth constrained peers.
type MessageBatch struct {
mu sync.Mutex
messages []*Message
}
func (mb *MessageBatch) add(msg *Message) {
mb.mu.Lock()
defer mb.mu.Unlock()
mb.messages = append(mb.messages, msg)
}
func (mb *MessageBatch) take() []*Message {
mb.mu.Lock()
defer mb.mu.Unlock()
messages := mb.messages
mb.messages = nil
return messages
}
type messageBatchAndPublishOptions struct {
messages []*Message
opts *BatchPublishOptions

View File

@ -1600,12 +1600,10 @@ func (p *PubSub) PublishBatch(batch *MessageBatch, opts ...BatchPubOpt) error {
setDefaultBatchPublishOptions(publishOptions)
p.sendMessageBatch <- messageBatchAndPublishOptions{
messages: batch.messages,
messages: batch.take(),
opts: publishOptions,
}
// Clear the batch's messages in case a user reuses the same batch object
batch.messages = nil
return nil
}

View File

@ -257,7 +257,7 @@ func (t *Topic) AddToBatch(ctx context.Context, batch *MessageBatch, data []byte
}
return err
}
batch.messages = append(batch.messages, msg)
batch.add(msg)
return nil
}