From 727d053dd4beb5fc97e4d255ffbd58bdd680f35d Mon Sep 17 00:00:00 2001 From: "Cesarini, Daniele" <daniele.cesarini@sky.uk> Date: Tue, 14 Nov 2017 15:48:14 +0000 Subject: [PATCH] Fixing #751 K8SApiServerLatency always triggering --- assets/prometheus/rules/kube-apiserver.rules.yaml | 4 ++-- manifests/grafana/grafana-dashboards.yaml | 1 + manifests/prometheus/prometheus-k8s-rules.yaml | 4 ++-- 3 files changed, 5 insertions(+), 4 deletions(-) diff --git a/assets/prometheus/rules/kube-apiserver.rules.yaml b/assets/prometheus/rules/kube-apiserver.rules.yaml index 55ebe025..50982b05 100644 --- a/assets/prometheus/rules/kube-apiserver.rules.yaml +++ b/assets/prometheus/rules/kube-apiserver.rules.yaml @@ -11,8 +11,8 @@ groups: disappeared from service discovery. summary: API server unreachable - alert: K8SApiServerLatency - expr: histogram_quantile(0.99, sum(apiserver_request_latencies_bucket{subresource!="log",verb!~"^(?:CONNECT|WATCHLIST|WATCH|PROXY)$"}) - WITHOUT (instance, resource)) / 1e+06 > 1 + expr: histogram_quantile(0.99, sum(rate(apiserver_request_latencies_bucket{subresource!="log",verb!~"^(?:CONNECT|WATCHLIST|WATCH|PROXY)$"}[10m])) + by (le)) / 1e+06 > 1 for: 10m labels: severity: warning diff --git a/manifests/grafana/grafana-dashboards.yaml b/manifests/grafana/grafana-dashboards.yaml index aff751da..a55a8df5 100644 --- a/manifests/grafana/grafana-dashboards.yaml +++ b/manifests/grafana/grafana-dashboards.yaml @@ -6287,3 +6287,4 @@ data: "type": "prometheus", "url": "http://prometheus-k8s.monitoring.svc:9090" } +--- diff --git a/manifests/prometheus/prometheus-k8s-rules.yaml b/manifests/prometheus/prometheus-k8s-rules.yaml index d470c7bf..6be3ee49 100644 --- a/manifests/prometheus/prometheus-k8s-rules.yaml +++ b/manifests/prometheus/prometheus-k8s-rules.yaml @@ -227,8 +227,8 @@ data: disappeared from service discovery. summary: API server unreachable - alert: K8SApiServerLatency - expr: histogram_quantile(0.99, sum(rate(apiserver_request_latencies_bucket{subresource!="log",verb!~"^(?:CONNECT|WATCHLIST|WATCH|PROXY)$"}[10m])) - by (le)) / 1e+06 > 1 + expr: histogram_quantile(0.99, sum(rate(apiserver_request_latencies_bucket{subresource!="log",verb!~"^(?:CONNECT|WATCHLIST|WATCH|PROXY)$"}[10m])) + by (le)) / 1e+06 > 1 for: 10m labels: severity: warning -- GitLab