Recovery and troubleshooting

What it covers

• General: preserve evidence first; do not delete state/journal/lock/audit merely to make it start.
• WEB failure: dctl-runtime/dctl-diagnostics → services/dependencies → logs/events → bundle → guided unit plan → health.
• Stale worker: heartbeat/threshold/dependency/load/schedule → refresh → planned restart if recommended.
• Incomplete transaction: process/lock/journal/backend log → controlled recovery → registry reconciliation.
• Remote offline: Last seen/address/firewall/Remote Administration/trust/fingerprint before re-pairing.
• Damaged state: secure persistence/restore staging/safe repair, not empty overwrite.

How to use it

1. Identify the domain.
2. Export evidence.
3. Check plan/impact.
4. Run one controlled action.
5. Verify health/close or review the finding.

Important rules

• Unknown impact/multiple core domains: stop and review the bundle.
• Guessing is not recovery.
• Do not mark unresolved state as success.