diff --git a/jsonnet/kube-prometheus/alerts/prometheus-operator.libsonnet b/jsonnet/kube-prometheus/alerts/prometheus-operator.libsonnet index a8ddd13affe7800b470e03e5412ce2b709d3e1d3..731994a2fba904c9961dab1d78a2b5f28ef68a2a 100644 --- a/jsonnet/kube-prometheus/alerts/prometheus-operator.libsonnet +++ b/jsonnet/kube-prometheus/alerts/prometheus-operator.libsonnet @@ -7,7 +7,7 @@ { alert: 'PrometheusOperatorListErrors', expr: ||| - (sum by (controller,namespace) (rate(prometheus_operator_list_operations_failed_total{%(prometheusOperatorSelector)s}[1h])) / sum by (controller,namespace) (rate(prometheus_operator_list_operations_total{%(prometheusOperatorSelector)s}[1h]))) > 0.4 + (sum by (controller,namespace) (rate(prometheus_operator_list_operations_failed_total{%(prometheusOperatorSelector)s}[10m])) / sum by (controller,namespace) (rate(prometheus_operator_list_operations_total{%(prometheusOperatorSelector)s}[10m]))) > 0.4 ||| % $._config, labels: { severity: 'warning', @@ -20,7 +20,7 @@ { alert: 'PrometheusOperatorWatchErrors', expr: ||| - (sum by (controller,namespace) (rate(prometheus_operator_watch_operations_failed_total{%(prometheusOperatorSelector)s}[1h])) / sum by (controller,namespace) (rate(prometheus_operator_watch_operations_total{%(prometheusOperatorSelector)s}[1h]))) > 0.4 + (sum by (controller,namespace) (rate(prometheus_operator_watch_operations_failed_total{%(prometheusOperatorSelector)s}[10m])) / sum by (controller,namespace) (rate(prometheus_operator_watch_operations_total{%(prometheusOperatorSelector)s}[10m]))) > 0.4 ||| % $._config, labels: { severity: 'warning', diff --git a/manifests/prometheus-rules.yaml b/manifests/prometheus-rules.yaml index 185c56d143a8e4b8921f0affb1bbcdf3d3d169a7..0c1eb32de647269a13a70b621eb76b5b6bd95eb4 100644 --- a/manifests/prometheus-rules.yaml +++ b/manifests/prometheus-rules.yaml @@ -1875,7 +1875,7 @@ spec: message: Errors while performing List operations in controller {{$labels.controller}} in {{$labels.namespace}} namespace. expr: | - (sum by (controller,namespace) (rate(prometheus_operator_list_operations_failed_total{job="prometheus-operator",namespace="monitoring"}[1h])) / sum by (controller,namespace) (rate(prometheus_operator_list_operations_total{job="prometheus-operator",namespace="monitoring"}[1h]))) > 0.4 + (sum by (controller,namespace) (rate(prometheus_operator_list_operations_failed_total{job="prometheus-operator",namespace="monitoring"}[10m])) / sum by (controller,namespace) (rate(prometheus_operator_list_operations_total{job="prometheus-operator",namespace="monitoring"}[10m]))) > 0.4 for: 15m labels: severity: warning @@ -1884,7 +1884,7 @@ spec: message: Errors while performing Watch operations in controller {{$labels.controller}} in {{$labels.namespace}} namespace. expr: | - (sum by (controller,namespace) (rate(prometheus_operator_watch_operations_failed_total{job="prometheus-operator",namespace="monitoring"}[1h])) / sum by (controller,namespace) (rate(prometheus_operator_watch_operations_total{job="prometheus-operator",namespace="monitoring"}[1h]))) > 0.4 + (sum by (controller,namespace) (rate(prometheus_operator_watch_operations_failed_total{job="prometheus-operator",namespace="monitoring"}[10m])) / sum by (controller,namespace) (rate(prometheus_operator_watch_operations_total{job="prometheus-operator",namespace="monitoring"}[10m]))) > 0.4 for: 15m labels: severity: warning