Skip to content

Modern Docker Management, Designed for Everyone

fix: auto update job not running

Kyle Mendell41d8a6596c83
diff --git a/backend/internal/bootstrap/jobs_bootstrap.go b/backend/internal/bootstrap/jobs_bootstrap.go
index 2dc4b404d..84db50858 100644
--- a/backend/internal/bootstrap/jobs_bootstrap.go
+++ b/backend/internal/bootstrap/jobs_bootstrap.go
@@ -92,6 +92,7 @@ func setupSettingsCallbacks(appServices *Services, appConfig *config.Config, ima
}
}
appServices.Settings.OnAutoUpdateSettingsChanged = func(ctx context.Context) {
+ slog.DebugContext(ctx, "AutoUpdateSettingsChanged callback triggered")
if err := autoUpdateJob.Reschedule(ctx); err != nil {
slog.WarnContext(ctx, "Failed to reschedule auto-update job", "error", err)
}
diff --git a/backend/internal/job/job.go b/backend/internal/job/job.go
index e532e0442..aaa151b37 100644
--- a/backend/internal/job/job.go
+++ b/backend/internal/job/job.go
@@ -26,6 +26,9 @@ func (s *Scheduler) Run(ctx context.Context) error {
slog.InfoContext(ctx, "Starting job scheduler")
s.scheduler.Start() // Start the scheduler, non-blocking
+ time.Sleep(100 * time.Millisecond)
+ s.LogJobStatus(ctx)
+
// Wait for the context to be done (e.g., application shutdown)
<-ctx.Done()
@@ -48,38 +51,39 @@ func (s *Scheduler) RegisterJob(
runImmediately bool,
) error {
jobOptions := []gocron.JobOption{
- gocron.WithContext(ctx),
gocron.WithName(name),
gocron.WithEventListeners(
gocron.BeforeJobRuns(func(jobID uuid.UUID, jobName string) {
- slog.InfoContext(ctx, "Job starting", "name", name, "id", jobID.String())
+ slog.Info("Job starting", "name", name, "id", jobID.String())
}),
gocron.AfterJobRuns(func(jobID uuid.UUID, jobName string) {
- slog.InfoContext(ctx, "Job finished successfully", "name", name, "id", jobID.String())
+ slog.Info("Job finished successfully", "name", name, "id", jobID.String())
}),
gocron.AfterJobRunsWithError(func(jobID uuid.UUID, jobName string, err error) {
- slog.ErrorContext(ctx, "Job failed", "name", name, "id", jobID.String(), "error", err)
+ slog.Error("Job failed", "name", name, "id", jobID.String(), "error", err)
}),
),
}
task := gocron.NewTask(func() {
if err := taskFunc(ctx); err != nil {
- slog.ErrorContext(ctx, "Error executing task function", "name", name, "error", err)
+ slog.Error("Error executing task function", "name", name, "error", err)
}
})
+ var job gocron.Job
var err error
if runImmediately {
- _, err = s.scheduler.NewJob(definition, task, append(jobOptions, gocron.WithStartAt(gocron.WithStartImmediately()))...)
+ job, err = s.scheduler.NewJob(definition, task, append(jobOptions, gocron.WithStartAt(gocron.WithStartImmediately()))...)
} else {
- _, err = s.scheduler.NewJob(definition, task, jobOptions...)
+ job, err = s.scheduler.NewJob(definition, task, jobOptions...)
}
if err != nil {
return fmt.Errorf("failed to register job %q: %w", name, err)
}
- slog.InfoContext(ctx, "Job registered successfully", "name", name)
+ nextRun, _ := job.NextRun()
+ slog.InfoContext(ctx, "Job registered successfully", "name", name, "nextRun", nextRun.Format(time.RFC3339), "runImmediately", runImmediately)
return nil
}
@@ -98,7 +102,21 @@ func (s *Scheduler) RescheduleDurationJobByName(
taskFunc func(ctx context.Context) error,
runImmediately bool,
) error {
+ slog.DebugContext(ctx, "Rescheduling job", "name", name, "interval", interval.String())
s.RemoveJobByName(name)
definition := gocron.DurationJob(interval)
return s.RegisterJob(ctx, name, definition, taskFunc, runImmediately)
}
+
+func (s *Scheduler) LogJobStatus(ctx context.Context) {
+ jobs := s.scheduler.Jobs()
+ slog.InfoContext(ctx, "Active jobs in scheduler", "count", len(jobs))
+ for _, j := range jobs {
+ nextRun, err := j.NextRun()
+ if err != nil {
+ slog.WarnContext(ctx, "Could not get next run for job", "name", j.Name(), "error", err)
+ continue
+ }
+ slog.InfoContext(ctx, "Job status", "name", j.Name(), "id", j.ID().String(), "nextRun", nextRun.Format(time.RFC3339))
+ }
+}
diff --git a/backend/internal/services/updater_service.go b/backend/internal/services/updater_service.go
index 3661b98b5..81ac43318 100644
--- a/backend/internal/services/updater_service.go
+++ b/backend/internal/services/updater_service.go
@@ -187,12 +187,18 @@ func (s *UpdaterService) ApplyPending(ctx context.Context, dryRun bool) (*update
host, repo, tag := s.parseNormalizedRef(normNew)
authHeader, _, _, _ := arcRegistry.ResolveAuthHeaderForRepository(ctx, host, repo, tag, enabledRegs)
check := digestChecker.CheckImageNeedsUpdate(ctx, normNew, authHeader)
+ skipPull := false
+
if check.CheckedViaAPI && check.Error == nil && !check.NeedsUpdate {
item.Status = "skipped"
item.Error = "image already up to date"
out.Skipped++
- out.Items = append(out.Items, item)
- _ = s.recordRun(ctx, item)
+ // We skip checking for pull, but we still proceed to container update checks
+ // treating this as "successful" for the pipeline, but invalidating oldIDs
+ // because they represent the *current* image, not a stale one.
+ plans[i].pulled = true
+ plans[i].oldIDs = nil
+ skipPull = true
s.logAutoUpdate(ctx, s.severityFromStatus(item.Status), models.JSON{
"phase": "image_pull",
@@ -204,34 +210,35 @@ func (s *UpdaterService) ApplyPending(ctx context.Context, dryRun bool) (*update
"checkedViaApi": true,
"error": item.Error,
})
- continue
}
- if err := s.imageService.PullImage(ctx, p.newRef, io.Discard, systemUser, nil); err != nil {
- item.Status = "failed"
- item.Error = err.Error()
- out.Failed++
- } else {
- item.Status = "updated"
- item.UpdateApplied = true
- out.Updated++
- plans[i].pulled = true
- for _, id := range p.oldIDs {
- if id != "" {
- oldIDSet[id] = struct{}{}
+ if !skipPull {
+ if err := s.imageService.PullImage(ctx, p.newRef, io.Discard, systemUser, nil); err != nil {
+ item.Status = "failed"
+ item.Error = err.Error()
+ out.Failed++
+ } else {
+ item.Status = "updated"
+ item.UpdateApplied = true
+ out.Updated++
+ plans[i].pulled = true
+ for _, id := range p.oldIDs {
+ if id != "" {
+ oldIDSet[id] = struct{}{}
+ }
}
}
+ s.logAutoUpdate(ctx, s.severityFromStatus(item.Status), models.JSON{
+ "phase": "image_pull",
+ "imageOld": p.oldRef,
+ "imageNew": p.newRef,
+ "status": item.Status,
+ "error": item.Error,
+ })
}
+
out.Items = append(out.Items, item)
_ = s.recordRun(ctx, item)
-
- s.logAutoUpdate(ctx, s.severityFromStatus(item.Status), models.JSON{
- "phase": "image_pull",
- "imageOld": p.oldRef,
- "imageNew": p.newRef,
- "status": item.Status,
- "error": item.Error,
- })
}
// Build maps for fast matching later (only for successfully pulled updates)
@@ -996,6 +1003,9 @@ func (s *UpdaterService) restartContainersUsingOldIDs(ctx context.Context, oldID
markedForRestart := map[string]bool{}
containersWithDeps := make([]arcaneupdater.ContainerWithDeps, 0, len(list))
+ // Cache resolved IDs for newRefs to avoid repeated API calls
+ targetImageIDs := map[string][]string{}
+
for _, c := range list {
inspect, err := dcli.ContainerInspect(ctx, c.ID)
if err != nil {
@@ -1040,6 +1050,25 @@ func (s *UpdaterService) restartContainersUsingOldIDs(ctx context.Context, oldID
}
}
+ if newRef != "" {
+ // Check if container is already on the target image
+ tids, cached := targetImageIDs[newRef]
+ if !cached {
+ tids, _ = s.resolveLocalImageIDsForRef(ctx, newRef)
+ targetImageIDs[newRef] = tids
+ }
+
+ for _, tid := range tids {
+ if tid == inspect.Image {
+ // Already on target image
+ slog.InfoContext(ctx, "restartContainersUsingOldIDs: container already on target image; skipping restart",
+ "containerId", c.ID, "containerName", dep.Name, "imageID", inspect.Image, "newRef", newRef)
+ newRef = ""
+ break
+ }
+ }
+ }
+
p := &restartPlan{cnt: c, inspect: inspect, newRef: newRef, match: match, explicit: newRef != ""}
plansByName[dep.Name] = p
if p.explicit {