Choose containment from a release and job-processing record.

Instruction: Compare the available actions, choose containment from the exhibits and define evidence required before exports resume.

Context:

All records are fictional UTC. The incident commander requests a decision at 09:08. Product permits temporary export unavailability; checkout and browsing should remain available.

Time / recordEvidence
09:00 release manifestv2 begins rollout; export_v2 flag enabled. v1 declares required column export_legacy_key; v2 does not require it.
09:03 migration receiptDROP COLUMN export_legacy_key completed. Restoring its values has not been tested.
09:05 endpoint metricsExport failures 15%; checkout and browsing match their pre-release baselines. Cause is not established.
09:07 containment rehearsalWith export_v2 disabled, UI and direct API return documented temporary-unavailable responses; enqueue count stays zero. Running jobs can pause at a durable checkpoint without committing partial output.
09:08 repair candidatev2.1 builds and unit tests pass. Representative export-output and resumed-job checks are pending.

Available controls are the verified flag, checkpoint pause, deployment stop, previous-artifact rollback and a candidate forward repair. Explain which are currently justified, which evidence would change your choice, and how to avoid calling containment full recovery.

Updated

Official answer available

Read the opening below, then unlock the full answer and practical guidance.

I would stop the rollout, disable new exports through the verified server-side flag and pause running jobs at their durable checkpoint...

Upgrade to view official answer

Related Questions