Victoria Metrics / Operator¶

Overview for operator VictoriaMetrics v0.25.0 or higher

Tags¶

Name	Description	Thresholds
Version	Victoria Metrics operator version.
CRD Objects count by controller	Number of objects at kubernetes cluster per each controller	Default: Mode: absolute Level 1: 80
Uptime	Victoria Metrics operator uptime.	Default: Mode: absolute Level 1: 80
Reconciliation rate by controller	Total number of reconciliations per controller.
Log message rate	Victoria Metrics operator log message rate.

Name	Description	Thresholds	Repeat
reconcile errors by controller	Non zero metrics indicates about error with CR object definition (typos or incorrect values) or errors with kubernetes API connection.
throttled reconcilation events	Operator limits number of reconcilation events to 5 events per 2 seconds. For now, this limit is applied only for vmalert and vmagent controllers. It should reduce load at kubernetes cluster and increase operator performance.
Wokring queue depth	Number of objects waiting in the queue for reconciliation. Non-zero values indicate that operator cannot process CR objects changes with the given resources.
Reconcilation latency by controller	For controllers with StatefulSet it's ok to see latency greater then 3 seconds. It could be vmalertmanager,vmcluster or vmagent in statefulMode. For other controllers, latency greater then 1 second may indicate issues with kubernetes cluster or operator's performance.

Name	Description	Thresholds	Repeat
Memory usage ($instance)	Victoria Metrics operator memory usage.
CPU ($instance)	Victoria Metrics operator CPU usage.
Goroutines ($instance)	Total number of goroutines.
GC duration ($instance)	Victoria Metrics operator avg GC duration.