helm: Keda prometheus queries are not mimir-specific about mimir HOT 3 CLOSED

We recommend running Mimir in its own namespace. Much of the helm chart, jsonnet, and mixin (dashboards and alerts) assumes that Mimir is in its own namespace as you found out

from mimir.

The dashboards, alerts use the cluster label to differentiate between the different deployments. I have configured our servicemonitors (kube-state-metrics, kubelet/cadvisor, etc) to add those labels. The keda scaledobjects could use them too.

The helm chart otherwise works fine as a subchart beside tempo, grafana-agent, etc, for our full monitoring stack.

from mimir.

the mixin assumes mimir is deployed in its own namespace. I think synchronising autoscaling and alerting and dashboards between products (tempo, loki, mimir) will make it hard because each manages changes on its own. I'm inclined to close this issue

from mimir.