Skip to main content
Stoneturner runs background jobs to keep your synced data up to date and recover from transient failures.
If self-hosting, set CRON_ENABLED=false in .env to disable all three cron jobs. Manual sync and retry endpoints still work.

​
Sync schedules

Each integration can have its own sync frequency. Configure it from the knowledge base UI — open the Sync frequency dialog (sync-schedule-dialog.tsx) and choose Daily, Weekly, Monthly, or No Schedule.

​
Retries

When a pipeline step fails, it writes a syncTask row with status: "FAILED", the step name, and inputs (cursor, offset, or other resume state). These steps are retried once per day until they reach a SUCCESS state OR have 3 FAILED attempts in a row.

​
Manual trigger

If you’d like to trigger a retry immediately, you can manually triggers the retry for all failed tasks in the UI.

​
Common questions

Select No Schedule in the sync frequency dialog. Stoneturner removes that integration’s syncPipeline row, so the midnight scheduler ignores it. You can still trigger syncs manually.
Yes. Use the per-integration sync button in the UI, which calls POST /api/sync/:integration for a full sync. For an incremental sync only, use POST /api/sync/updates/:integration.
Check the server console. retry-cron.ts logs each failed retry with the integration, step name, task ID, and error message.
Deleting an integration’s credential calls DELETE /api/sync/:integration, which invokes that integration’s deleteSync implementation. This removes its pipeline row, purges artifacts and embeddings, and clears integration-specific data.
For a deeper look at how sync steps record progress and resume state, see Add integrations and the architecture page.