easylinux/kube-prometheus
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge pull request #635 from lilic/cherry-pick-alerts

Browse Source 
[release-0.6] jsonnet/prometheus-operator.libsonnet: Adjust alerts range
This commit is contained in:
Frederic Branczyk
2020-08-06 15:40:04 +02:00
committed by GitHub
parent 85e22303f0 09a305bc0e
commit f1d92c8a80
2 changed files with 4 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
jsonnet/kube-prometheus/alerts/prometheus-operator.libsonnet
View File

@@ -7,7 +7,7 @@
{ {
alert: 'PrometheusOperatorListErrors', alert: 'PrometheusOperatorListErrors',
expr: ||| 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, ||| % $._config,
labels: { labels: {
severity: 'warning', severity: 'warning',
@@ -20,7 +20,7 @@
{ {
alert: 'PrometheusOperatorWatchErrors', alert: 'PrometheusOperatorWatchErrors',
expr: ||| 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, ||| % $._config,
labels: { labels: {
severity: 'warning', severity: 'warning',

4
manifests/prometheus-rules.yaml
View File

@@ -1875,7 +1875,7 @@ spec:
message: Errors while performing List operations in controller {{$labels.controller}} message: Errors while performing List operations in controller {{$labels.controller}}
in {{$labels.namespace}} namespace. in {{$labels.namespace}} namespace.
expr: | 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 for: 15m
labels: labels:
severity: warning severity: warning
@@ -1884,7 +1884,7 @@ spec:
message: Errors while performing Watch operations in controller {{$labels.controller}} message: Errors while performing Watch operations in controller {{$labels.controller}}
in {{$labels.namespace}} namespace. in {{$labels.namespace}} namespace.
expr: | 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 for: 15m
labels: labels:
severity: warning severity: warning