strimzi-kafka-operator
strimzi-kafka-operator copied to clipboard
Consistent colours for pods in Grafana
Describe the bug
The provided grafana dashboards don't use a consistent colour across the charts for the same zookeeper node or kafka broker. This can make reading and correlating the charts more difficult than it needs to be. See https://github.com/grafana/grafana/issues/6662
To Reproduce Steps to reproduce the behavior:
- Just look at the default grafana dashboard provided for a ZK cluster with ≥ 2 nodes.
Expected behavior
You'll see zookeeper-n has different colours in the different timeseries charts. So cyan might be zookeeper-0 in one chart but zookeeper-2 on another.
Triaged on 2.8.2022: This is still an issue in the current dashboards.
I did some digging around and I think the only thing we can do is to update prometheus queries to return one value per pod name. I was able to reach a sufficient results with this:
sum(zookeeper_outstandingrequests{namespace="strimzi-kafka",strimzi_io_cluster="heimdall",strimzi_io_kind="Kafka",kubernetes_pod_name=~"heimdall-.*"}) by (pod)
Only Y-axis has different scale, but graph values are the same with this query