Open ripa1993 opened 1 day ago
In the built mixins, the metrics assume pod names starting with
mimir-[component]
but in mimir-distributed chart the components are namedmimir-distributed-[component]
I think, with Helm chart this depends on the release name, user passes to the helm install
command. That is, following Mimir Helm chart's getting started guide (note the "mimir" release name in the command above):
% helm install mimir grafana/mimir-distributed --namespace test-monitoring
This installs k8s resources with the mimir-
prefix, i.e.
% kubectl -n test-monitoring get po
NAME READY STATUS RESTARTS AGE
mimir-compactor-0 1/1 Running 0 8m27s
mimir-store-gateway-zone-c-0 1/1 Running 0 8m27s
mimir-store-gateway-zone-b-0 1/1 Running 0 8m26s
mimir-store-gateway-zone-a-0 1/1 Running 0 8m27s
mimir-ingester-zone-c-0 1/1 Running 0 8m27s
mimir-ingester-zone-b-0 1/1 Running 0 8m27s
mimir-ingester-zone-a-0 1/1 Running 0 8m26s
···
The same is true for the helm template
command, some users use to deploy Mimir. The dashboards expect "mimir" template name:
% helm template grafana/mimir-distributed --name-template mimir
If you chose to use another release name, for the moment, you need to rebuild the dashboards yourself, following the guide.
Describe the bug
Grafana dashboards delivered with the Helm Chart use metrics that reference wrong pod names. In the built mixins, the metrics assume pod names starting with
mimir-[component]
but in mimir-distributed chart the components are namedmimir-distributed-[component]
To Reproduce
Steps to reproduce the behavior:
Expected behavior
Dashboards panel should be able to fetch the correct data.
Environment
Additional Context