CODEWITHSUNDEEP
kubernetesapache-kafkakubernetes-helmgrafanametrics
gstackoverflow
gstackoverflow

Reputation: 37034

How to monitor kafka system metrics like CPU load and memory usage?

I've enabled both exporters:

chart.yaml:

...
dependencies:
  ...
  - name: kafka
    version: 17.1.0
    repository: "@bitnami"
    condition: kafka.enabled
  ...

values.yaml:

kafka:
  metrics:
    kafka:
      enabled: true
    jmx:
      enabled: true
    serviceMonitor:
      enabled: true
      labels:
        my.custom.label/service-monitor: "1.0"

I see that both of them are appeared on status->Service Discovery page of Prometheus so I consider them working.

Next step: I add following grafana dashboard: https://grafana.com/grafana/dashboards/12483

But most of metrics(not all) are not available on that dashboard. enter image description here For example one of the panel uses jvm_memory_bytes_used metric but I don't see this metric on prometheus side

How can I fix it ?

Upvotes: 0

Views: 1293

Answers (1)

OneCricketeer
OneCricketeer

Reputation: 191710

CPU and memory load would need to come from a different exporter (e.g. node_exporter) or for related pod/container metrics within k8s, then cAdvisor.

For example one of the panel uses jvm_memory_bytes_used metric but I don't see this metric on prometheus side

Unclear what Datasource your Grafana dashboard is using. If it is Prometheus, then you will want to first ensure you are querying within the appropriate time range. But only after you manually grep the metrics endpoint that is being scraped to ensure your JMX exporter is actually reporting that data.

Upvotes: 1

Related Questions