Guardian adds candidate deployment and validation before a new configuration is promoted. It complements the application’s update strategy and requires suitable checks.
Configure and release
Prepare a meaningful XHealth check, enable Guardian for a supported application, and review resources for the candidate as well as the current application.
With Guardian enabled, changes to the image, container runtime options, XConfig, or group trigger candidate validation. Group changes are available through the API. Local-volume stateful applications do not support Guardian.
After saving a change, follow the release phase on the application overview.
Phases and decisions
| Phase | What to inspect |
|---|---|
| Deploying | Candidate scheduling, image pull, and startup |
| Validating | Check progress and assertions |
| Failed | Candidate or check failure details |
| Promoting | Replacement of the live configuration |
When the interface offers retry, skip, or cancel, choose according to the observed failure and current phase. Skipping accepts the change without successful validation; it does not turn a failed check into a passing one. Cancellation availability depends on the release state.
Diagnose and verify
Use the separate Guardian application and Guardian check log sources to identify which container failed. Check target paths, image tools, environment, candidate capacity, and assertions.
After promotion, verify live replicas and the actual business endpoint.
Review update strategies and rollback before relying on automatic recovery.