Merge pull request #1505 from mapellidario/20240612_crabserver_openfds

crabserver - lower threshold for pod restart with too many open fds
dmwm · Jul 19, 2024 · 0cb2d75 · 0cb2d75
2 parents 17f07fc + 13b6680
commit 0cb2d75
Show file tree

Hide file tree

Showing 2 changed files with 7 additions and 40 deletions.
diff --git a/kubernetes/cmsweb/monitoring/prometheus/rules/crabserver.rules b/kubernetes/cmsweb/monitoring/prometheus/rules/crabserver.rules
@@ -1,10 +1,8 @@
 groups:
 - name: crabserver
   rules:
-  - record: avg_open_fds_10m
-    expr: avg_over_time(crabserver_process_open_fds[10m])
-  - record: avg_open_fds_30m
-    expr: avg_over_time(crabserver_process_open_fds[30m])
+  - record: avg_open_fds_8m
+    expr: avg_over_time(crabserver_process_open_fds[8m])
   - alert: CRAB server is down
     expr: crabserver_num_cpus == 0
     for: 5m
@@ -17,20 +15,8 @@ groups:
     annotations:
       summary: "crabserver {{ $labels.env }} is down"
       description: "{{ $labels.env }} has been down for more than 5m"
-  - alert: CRAB server service has large number of fds
-    expr: avg_open_fds_10m > 75
-    for: 1m
-    labels:
-      severity: warning
-      tag: cmsweb
-      service: crab
-      host: "{{ $labels.host }}"
-      action: Please check CRAB server on {{ $labels.instance }} and possibly restart it
-    annotations:
-      summary: "CRAB {{ $labels.env }} environment"
-      description: "{{ $labels.env }} has large level of fds {{ $value }} (avg 10m) for more than 1m"
   - alert: CRAB server service has high number of fds
-    expr: avg_open_fds_30m > 100
+    expr: avg_open_fds_8m > 50
     for: 1m
     labels:
       severity: high
@@ -40,4 +26,4 @@ groups:
       action: Please restart CRAB server on {{ $labels.instance }}
     annotations:
       summary: "CRAB {{ $labels.env }} environment"
-      description: "{{ $labels.env }} has high level of fds {{ $value }} (avg 30m) for more than 1m"
+      description: "{{ $labels.env }} has high level of fds {{ $value }} (avg 8m) for more than 1m"
diff --git a/kubernetes/cmsweb/monitoring/prometheus/rules/crabserver.test b/kubernetes/cmsweb/monitoring/prometheus/rules/crabserver.test
@@ -25,27 +25,8 @@ tests:
                  description: "prod has been down for more than 5m"
 - interval: 1m
   input_series:
-  - series: 'avg_open_fds_10m{env="prod",instance="test-instance",host="k8s-test"}'
-    values: '76+1x100'
-  alert_rule_test:
-      - eval_time: 10m
-        alertname: CRAB server service has large number of fds
-        exp_alerts:
-            - exp_labels:
-                 severity: warning
-                 tag: cmsweb
-                 service: crab
-                 host: k8s-test
-                 action: Please check CRAB server on test-instance and possibly restart it
-                 instance: test-instance
-                 env: prod
-              exp_annotations:
-                 summary: "CRAB prod environment"
-                 description: "prod has large level of fds 86 (avg 10m) for more than 1m"
-- interval: 1m
-  input_series:
-  - series: 'avg_open_fds_30m{env="prod",instance="test-instance",host="k8s-test"}'
-    values: '101+1x100'
+  - series: 'avg_open_fds_8m{env="prod",instance="test-instance",host="k8s-test"}'
+    values: '51+1x100'
   alert_rule_test:
       - eval_time: 10m
         alertname: CRAB server service has high number of fds
@@ -60,4 +41,4 @@ tests:
                  env: prod
               exp_annotations:
                  summary: "CRAB prod environment"
-                 description: "prod has high level of fds 111 (avg 30m) for more than 1m"
+                 description: "prod has high level of fds 61 (avg 8m) for more than 1m"