Maintenance mode runbook
Maintenance mode is an explicit operational state for planned or emergency work. It should not be used as a generic response to every isolated bug.
Before planned maintenance
- Define affected components and expected user impact.
- Schedule the maintenance window.
- Prepare rollback or abort criteria.
- Confirm the operator permissions needed.
- Communicate planned impact through the supported status/user channels.
Enable maintenance
Use the Control Center configuration surface when the current deployment supports maintenance mode.
Confirm whether the setting affects:
- web workspace
- Developer Platform
- background runs/builds
- new sign-ins
- all components or only selected surfaces
Do not assume one maintenance toggle safely pauses every subsystem.
During maintenance
Monitor:
- health checks
- database/cache/queue dependencies
- deployment/migration progress
- background jobs
- error rate
- user/API traffic behavior
Avoid starting unrelated configuration changes while the platform is intentionally in a reduced state.
Exit maintenance
- Confirm the change/migration completed.
- Run health checks.
- Verify representative user and API workflows.
- Re-enable normal traffic/access.
- Confirm error and latency metrics remain healthy.
- Close or update the associated incident/maintenance communication.
Emergency maintenance
For active platform degradation, maintenance mode can be a mitigation—not the final incident resolution. Keep the incident timeline and root-cause investigation separate.
Always verify maintenance mode is actually cleared in the intended environment. A successful deployment does not automatically restore traffic if the maintenance setting remains active.
