Metrics¶
The documentation status¶
In progress
Terms¶
Metric stability levels¶
| Level | Description |
|---|---|
| Alpha | Experimental. May change or be removed without notice. Not recommended for critical alerting. |
| Beta | Tested and functional. May change based on operational feedback. Suitable for dashboards and non-critical alerts. |
| Stable | Proven in production. Unlikely to change. Safe for alerting and SLO tracking. |
Metric sources¶
| Source | Description |
|---|---|
| Internal | Exposed by the iCRI process itself (port 13332). |
| External | Collected outside iCRI (e.g., node-exporter textfile collector, cron jobs). |
External metrics¶
icri_launchd_runs¶
| Stability | Beta |
| Type | Gauge |
| Source | External (node-exporter textfile collector) |
| File | /opt/node-exporter/textfile/icri_launchd_runs.prom |
Number of times launchd spawned the iCRI process since the last deploy. Resets to 1 on deploy (Ansible reloaded performs unload/load of the plist).
A value of 1 is normal. A value >1 indicates iCRI has crashed and been restarted by launchd.
Collection: A cron job (every minute, as root) parses the runs field from launchctl print system/com.jetbrains.icri.
Used in: - Alert icriHasRestarted: delta(icri_launchd_runs[15m]) > 0 - Dashboard: iCRI macOS Buildfarm, iCRI restarts panel
icri_launchd_last_exit_code¶
| Stability | Beta |
| Type | Gauge |
| Source | External (node-exporter textfile collector) |
| File | /opt/node-exporter/textfile/icri_launchd_last_exit_code.prom |
Exit code from the last iCRI process termination. Not emitted if the process has never exited since the last deploy (the .prom file is empty).
| Value | Meaning |
|---|---|
| 2 | Go panic |
| 9 | SIGKILL |
| 0 | Clean shutdown |
| absent | Never exited since last deploy |
TODO: add extra values or a reference to the exit code list.
Collection: A cron job (every minute, as root) parses the last exit code field from launchctl print system/com.jetbrains.icri. Only emitted when the value is numeric.
Used in: - Alert icriHasRestarted: included in alert summary annotation - Dashboard: iCRI macOS Buildfarm, iCRI restarts panel