Resiliency

All Data Jobs should be atomic and idempotent - it is OK for a job to fail somewhere in the middle or execution; subsequent restart should not cause data corruption.

Once deployed (if operators of Control Service have enabled it), Data Jobs will be automatically restarted on errors caused by infrastructure issues (e.g., intermittent issues with network connectivity).

➡️ Next Section: Data Jobs Development Workflow