Runbook: Agent Offline / Unresponsive
Runbook: Agent Offline / Unresponsive
Section titled “Runbook: Agent Offline / Unresponsive”Trigger: A customer AI agent ceases heartbeat pings, returns continuous 5xx errors, or fails execution tasks.
Step 1: Diagnose Connectivity & API Health
Section titled “Step 1: Diagnose Connectivity & API Health”- Check if
https://api.salitech.de/healthis returning healthy status. - Verify rate limit status and authentication logs for the given agent UUID.
Step 2: Check Agent Host / Process
Section titled “Step 2: Check Agent Host / Process”- Check agent process memory and CPU consumption.
- Inspect agent execution error logs.
- Validate API token validity and expiration.
Step 3: Mitigation
Section titled “Step 3: Mitigation”- Restart agent process / container.
- If token is invalid or expired, issue a replacement token following Runbook: Rotate API Key.
- Confirm telemetry handshake succeeds.