Operations Shared (all languages) active http-servicebatch-job

Retries are bounded and poison messages are quarantined

ops-016

Intent

Prevent one permanently bad message or operation from causing endless retry loops and queue starvation.

Applicability

Applies when the diff changes asynchronous consumers, queue workers, or background retry behavior.

What to inspect

Retry settings, dead-letter routing, replay tooling, and failure handlers.

Pass criteria

Retries are bounded and exhausted work goes to a dead-letter or quarantine path that operators can inspect or replay.

Fail criteria

The diff introduces unbounded retry, immediate requeue-on-failure loops, or no sink for terminally failing messages.

Do not flag

Synchronous one-shot retries or frameworks whose checked-in config already enforces retry bounds and dead-letter routing.

Confidence guidance

HIGH when retry loops or broker settings are explicit. MEDIUM when framework behavior is inferred. LOW when broker policy lives outside the repo.

Remediation

Set a max retry count and route exhausted items to a quarantine path.

Pass example

consumer:
  maxAttempts: 5
  deadLetterTopic: orders.failed

Fail example

consumer:
  requeueOnFailure: true
  maxAttempts: -1