BCAI/easyai-ai-gateway
3
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Fix realtime queued task counts

Browse Source
This commit is contained in:
wangbo
2026-05-12 10:41:07 +08:00
parent 2a91b31d12
commit 9ea83be718
6 changed files with 128 additions and 22 deletions
Download Patch File Download Diff File Expand all files Collapse all files
apps/api/internal/runner/service.go
+7 -5
View File
@@ -121,11 +121,13 @@ func (s *Service) execute(ctx context.Context, task store.GatewayTask, user *aut
maxFailoverDuration := maxFailoverDurationForCandidates(candidates, runnerPolicy)
attemptNo := task.AttemptCount
var lastErr error
var lastCandidate store.RuntimeModelCandidate
candidatesLoop:
for index, candidate := range candidates {
if index >= maxPlatforms {
break
}
lastCandidate = candidate
clientAttempts := clientAttemptsForCandidate(candidate)
var candidateErr error
for clientAttempt := 1; clientAttempt <= clientAttempts; clientAttempt++ {
@@ -180,7 +182,7 @@ candidatesLoop:
lastErr = err
candidateErr = err
if task.AsyncMode && store.RateLimitRetryable(err) {
queued, delay, queueErr := s.requeueRateLimitedTask(ctx, task, err)
queued, delay, queueErr := s.requeueRateLimitedTask(ctx, task, err, candidate)
if queueErr != nil {
return Result{}, queueErr
}
@@ -297,7 +299,7 @@ candidatesLoop:
return Result{Task: queued, Output: queued.Result}, &TaskQueuedError{Delay: 0}
}
if task.AsyncMode && errors.Is(lastErr, store.ErrRateLimited) && store.RateLimitRetryable(lastErr) {
queued, delay, queueErr := s.requeueRateLimitedTask(ctx, task, lastErr)
queued, delay, queueErr := s.requeueRateLimitedTask(ctx, task, lastErr, lastCandidate)
if queueErr != nil {
return Result{}, queueErr
}
@@ -525,12 +527,12 @@ func (s *Service) failTask(ctx context.Context, taskID string, code string, mess
return failed, nil
}
func (s *Service) requeueRateLimitedTask(ctx context.Context, task store.GatewayTask, cause error) (store.GatewayTask, time.Duration, error) {
func (s *Service) requeueRateLimitedTask(ctx context.Context, task store.GatewayTask, cause error, candidate store.RuntimeModelCandidate) (store.GatewayTask, time.Duration, error) {
delay := localRateLimitRetryAfter(cause)
if delay <= 0 {
delay = 5 * time.Second
}
queued, err := s.store.RequeueTask(ctx, task.ID, delay)
queued, err := s.store.RequeueTask(ctx, task.ID, delay, candidate.QueueKey)
if err != nil {
return store.GatewayTask{}, 0, err
}
@@ -546,7 +548,7 @@ func (s *Service) requeueRateLimitedTask(ctx context.Context, task store.Gateway
}
func (s *Service) requeueInterruptedAsyncTask(ctx context.Context, task store.GatewayTask) (store.GatewayTask, error) {
queued, err := s.store.RequeueTask(ctx, task.ID, 0)
queued, err := s.store.RequeueTask(ctx, task.ID, 0, "")
if err != nil {
return store.GatewayTask{}, err
}