Skip to content

SOP & Recovery

No recovery command should be run based solely on this generic template. Confirm target environment, backups and current process state first.

Incident response checklist

  1. Identify service and affected date window.
  2. Preserve sanitized error output and stable transaction IDs.
  3. Determine whether a live writer or reconciliation process is active.
  4. Verify API source of truth and Sheet schema/cursor positions.
  5. Perform a dry run for repair/backfill whenever supported.
  6. Confirm idempotency and lock coverage before any write.
  7. Reconcile again and record the result; do not infer success from a script exit code alone.
  8. Update issue, changelog and postmortem.

Rollback requirements

Every production-changing procedure should list owner, preconditions, backups, post-checks and a tested rollback path.