perf(actions): debounce runner heartbeat writes and throttle task picks (#38281)

3 reductions in the DB load generated by many runners polling
`FetchTask`:

**1. Debounce runner heartbeat writes**
Every poll wrote `last_online`, and every `UpdateTask`/`UpdateLog` wrote
`last_active` — while a runner streams logs that is many writes per
second per runner. These are now persisted only when stale enough to
actually affect the active/offline status (`ShouldPersistLastOnline` /
`ShouldPersistLastActive`), using the existing columns.

**2. Throttle concurrent task picks**
A new in-process semaphore (`MAX_CONCURRENT_TASK_PICKS`) bounds how many
runners run the task-assignment transaction at once, so a fleet polling
together cannot stampede the query. Throttled polls retry on their next
poll without advancing the runner's tasks version.

**3. Paginate the task-pick query**
`CreateTaskForRunner` previously loaded every waiting job in the
runner's scope into memory on each poll (no `LIMIT`). Now it pages
through the waiting backlog oldest-first with `LIMIT`, claiming the
first label-matching job.

---------

Co-authored-by: Zettat123 <zettat123@gmail.com>
This commit is contained in:
bircni
2026-07-07 21:16:20 +02:00
committed by GitHub
parent 97078b96cf
commit 308a6f12ae
10 changed files with 317 additions and 30 deletions
+7 -1
View File
@@ -194,9 +194,15 @@ func (s *Service) FetchTask(
// if the task version in request is not equal to the version in db,
// it means there may still be some tasks that haven't been assigned.
// try to pick a task for the runner that send the request.
if t, ok, err := actions_service.PickTask(ctx, freshRunner); err != nil {
if t, ok, throttled, err := actions_service.TryPickTask(ctx, freshRunner); err != nil {
log.Error("pick task failed: %v", err)
return nil, status.Errorf(codes.Internal, "pick task: %v", err)
} else if throttled {
// Concurrency limit reached: don't advance the runner's tasks version,
// so it retries on its next poll instead of sleeping until the next bump.
latestVersion = tasksVersion
// A steady stream here means MAX_CONCURRENT_TASK_PICKS is too low for the fleet.
log.Debug("task pick throttled for runner %q (id %d); it will retry on its next poll", freshRunner.Name, freshRunner.ID)
} else if ok {
task = t
}