Oskar Kapala
fb7828b52b
When a service becomes healthy (node-agent emits service_healthy → observer updates services.json), any previously queued redeploy/container_restart action is stale. Without cleanup, the queue accumulates old actions that require manual rejection. _cancel_resolved_pending_actions() runs after each reconcile cycle: - Reads all pending/*.json with type=redeploy or container_restart - If the service is now healthy in actual_state, moves action to cancelled/ with reason=drift_resolved_auto - Only pending actions are touched; approved/running are left to the operator Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> |
||
|---|---|---|
| .. | ||
| agent-system | ||
| control-plane | ||
| forgejo | ||
| mosquitto | ||
| node-agent | ||
| npm | ||
| ollama | ||
| stability-agent | ||
| zigbee2mqtt | ||
| .gitkeep | ||