[kube-prometheus-stack] fix API server error budget label

[Zakharden]

What this PR does / why we need it

Fixes the API server dashboard's 30-day error budget panel text so the rendered dashboard no longer describes the target as 0.990% availability. The panel now uses the same 99.000% scale as the title, adds the missing space in Error Budget, and uses a readable legend label.

Which issue this PR fixes

• fixes [kube-prometheus-stack] Tooltip for SLO Error Budget is off by factor of 100 #6173

Special notes for your reviewer

This is a focused chart-side correction for the issue reported against the rendered kube-prometheus-stack dashboard. I noticed the dashboard is generated from upstream mixins; if maintainers prefer an upstream-first fix, I can move the wording change there instead.

Validation

• helm template test charts/kube-prometheus-stack --kube-version 1.35.4 --show-only templates/grafana/dashboards-1.14/apiserver.yaml --set grafana.enabled=true --set kubeApiServer.enabled=true > /tmp/kps-apiserver-dashboard.yaml
• rg -n "0\\.990%|ErrorBudget|errorbudget" /tmp/kps-apiserver-dashboard.yaml charts/kube-prometheus-stack/templates/grafana/dashboards-1.14/apiserver.yaml -S (no matches)
• helm lint charts/kube-prometheus-stack --kube-version 1.35.4
• git diff --check

Checklist

• DCO signed
• Chart Version bumped
• Title of the PR starts with chart name (e.g. [prometheus-couchdb-exporter])

[jkroepke]

Need to be fixed upstream. See line header.

[Zakharden]

Thanks, moved the fix upstream: kubernetes-monitoring/kubernetes-mixin#1219.

I will not push additional direct edits to the generated dashboard here. Once the upstream change lands and is synced through kube-prometheus, this PR can be updated from the generated output.

[Zakharden]

Hi maintainers, could you please advise on the preferred path here?

I also opened kubernetes-monitoring/kubernetes-mixin#1219 with the upstream mixin fix for the same API server error budget wording.

If you prefer upstream-first, I can leave this PR on hold or close it once the generated dashboard is updated from the mixin. If a chart-side fix is still useful, I’d appreciate a review. Thanks.
@jkroepke

[QuentinBisson]

Yes this is the preferred way

[Zakharden]

Thanks for confirming. I will close this chart-side PR in favor of the upstream-first path through kubernetes-monitoring/kubernetes-mixin#1219.

The CI failure here is expected for a direct generated dashboard edit: ct lint runs the kube-prometheus-stack lint script, which re-syncs generated dashboards and rejects dashboard diffs that do not come from the upstream source. Once the upstream mixin change is accepted and synced downstream, the generated chart output can be updated through the normal sync path.
@QuentinBisson

[Zakharden]

Closing this chart-side PR now that the upstream-first fix has merged in kubernetes-monitoring/kubernetes-mixin#1219.

The direct generated-dashboard edit here is intentionally superseded by the upstream source change, so the kube-prometheus-stack generated dashboard can be updated through the normal sync path.