Rollback

Use this guide when a deployment wave causes service degradation or policy enforcement instability.

Rollback triggers

• Elevated block or step-up error budgets.
• Fleet-wide check-in failures.
• Unexpected configuration drift spikes.

Rollback procedure

1. Reapply the previous known-good runtime configuration profile.
2. Redeploy the previous signed package version if needed.
3. Validate endpoint recovery.

thoth health --json

4. Confirm Fleet endpoint status is healthy and current.
5. Keep incident tracking open until stability criteria are met.

Post-rollback validation

• Endpoint check-ins are healthy within 60 seconds.
• config_drifted=false for the rollback cohort.
• Install and runtime logs show normal startup.