search

ppgravitee / issuesforjiramig

Gravitee.io - API Platform - Issues
0 stars 0 forks source link

[APIM|GATEWAY] Node_health probe memory and cpu does not work correctly (community) #37

Open ppgravitee opened 2 years ago

ppgravitee commented 2 years ago

:collision: Describe the bug

Gravity was updated from version 3.15.3->3.17.3. We had an alert set up for node_health metrics I know that after the update, node_health has been changed Before(3.15.3) 0 - everything is ok Now(3.17.3) 1 - everything is ok We have changed the alert: Before(3.15.3) node_health != 0 Now(3.17.3) node_health == 0 The jump on the chart is the moment of the update. After the update, alerts began to come only for probe memory and cpu. But this does not reflect the real situation, memory is not used by 80%, we have not changed the default setting.

:sunrise_over_mountains: To Reproduce

Steps to reproduce the behaviour:

  1. Set gravitee_services_metrics_enabled=true, gravitee_services_metrics_prometheus_enabled=true
  2. Add a target to prometheus
  3. Wait for a while and execute the request in prometheus node_health{probe=~"memory|cpu"}
  4. See error

:rainbow: Expected behaviour

Node health probe=memory|cpu, if memory is not used more than the threshold value, the metric should be 1

Current behaviour

Node_health probe=memory|cpu does not display the real situation

:computer: Desktop:

Please complete the following information:

sync-by-unito[bot] commented 2 years ago

➤ Pierre-Philippe Beauchemin commented:

comment