A permits Lakeflow Job must not run the fee-publish task after a failed extract. What should be configured?

Job-level on-failure behavior must prevent municipal downstream tasks from running after upstream failures. A permits Lakeflow Job whose fee-publish task depends on a successful extract should define on-failure (or dependency) rules so fee publish does not run when extract fails. That protects residents from incorrect fee postings and keeps financial gold consistent with what was actually extracted.

Configuring task notifications so downstream fee publish does not run after a failed extract fails by concept: notifications only alert on extract state and do not enforce a success dependency, so fee publish could still run after a failed extract. Setting a retry policy with backoff so downstream fee publish does not run after a failed extract does not satisfy the requirement because retries address extract reliability, not whether fee publish runs after an extract failure. Configuring queueing and concurrency so downstream fee publish does not run after a failed extract does not create a success dependency, so fee publish could still run without a successful extract.

Exam caveat: some designs use repair runs or partial publish with explicit idempotent keys; the exam focus is that failed extract must not blindly trigger fee publish. Operational check: force an extract failure in staging, confirm fee-publish is skipped or blocked, verify no new fee gold rows appear for that run, and alert permits finance on the extract failure rather than on a silent bad publish.