Agent Status
The Agent Status tab on the Server Detail page shows the health of the PMP4PG agent running on the monitored server. Use this tab to verify that the agent is operating correctly and all collectors are active.
{screenshot: agent-status-tab}
Agent Information
| Field | Description |
|---|---|
| Agent ID | Unique identifier assigned at registration |
| Agent Version | Installed version of the PMP4PG agent binary |
| Status | Current agent status (ACTIVE, INACTIVE, ERROR) |
| Last Heartbeat | Timestamp of the most recent heartbeat received |
| Registered At | Date and time the agent was first registered |
| Operating System | OS of the monitored server (detected at registration) |
{screenshot: agent-info-panel}
Heartbeat Status
The heartbeat is sent by the agent every 30 seconds. The platform uses the heartbeat to determine whether the agent is alive.
| Last Heartbeat Age | Status |
|---|---|
| < 1 minute | 🟢 ACTIVE — agent is healthy |
| 1–5 minutes | 🟡 WARNING — agent may be starting or briefly interrupted |
| > 5 minutes | 🔴 INACTIVE — agent is likely down |
Collector Status
The Collectors section lists all configured collectors and their current state:
| Collector | Interval | Last Run | Status |
|---|---|---|---|
| ASH Samples | 2s | timestamp | ✅ Running |
| Heartbeat | 30s | timestamp | ✅ Running |
| pg_stat_statements | 60s | timestamp | ✅ Running |
| pg_stat_database | 60s | timestamp | ✅ Running |
| Database Size | 10m | timestamp | ✅ Running |
| OS Metrics | 60s | timestamp | ✅ Running |
| Server Snapshot | 30m | timestamp | ✅ Running |
{screenshot: collector-status-table}
A collector showing ❌ or a stale Last Run timestamp indicates a collection issue. Check the agent logs for details:
journalctl -u pmp4pg-agent -n 100 --no-pager
Agent Actions
From the Agent Status tab, an administrator can:
- Restart the agent — Sends a signal to the agent to restart (if supported by the deployment)
- View raw agent logs — Link to the log viewer (if configured)
{screenshot: agent-action-buttons}
Agent restart from the UI requires the agent to be running and able to receive control signals from the backend. If the agent is fully unresponsive, restart the service directly on the host:
systemctl restart pmp4pg-agent