Infrastructure
Status transitions
The status transitions card provides a detailed, time-ordered log of every status change across your fleet. Use it to understand exactly when agents went offline, what triggered the change, and what error accompanied it.
Transition log
Each row in the log represents a single status change event. The summary columns show the timestamp, agent name, and the transition direction (old status → new status). Transitions are listed in reverse chronological order so the most recent events appear first.
Filters
Two filters narrow the log to events you care about. The reason filter limits events by what caused the transition (heartbeat timeout, manual change, error detection, or cron health check). The target status filter limits events to transitions that ended in a specific status (online, idle, or offline).
Common filter combinations:
- Reason: heartbeat timeout + Status: offline — shows every time an agent went offline because it stopped sending heartbeats
- Reason: manual change — shows every operator-initiated status change
- Status: online — shows every recovery event
Expandable rows
Click any row to expand it and see the full details of the transition event.
| Field | Description |
|---|---|
| Timestamp | Exact time of the status change in your local timezone |
| Agent | Agent name and permanent ID |
| Old → New status | The status before and after the transition |
| Reason | Color-coded label: heartbeat timeout, manual change, error detection, or cron health check |
| Trigger source | What initiated the check: cron job, API call, user action, or proxy relay |
| HTTP status | HTTP status code from the heartbeat or health check that triggered the change |
| Latency | Response time of the health check in milliseconds |
| Error message | Error details if the transition was caused by a failure |
CSV export
Export the filtered transition log as a CSV file. The export includes all matching events (not just the current page) with all expanded-row fields.
Next
View all platform-wide state changes in one audit log. See Platform activity →