CFP: Support cilium metrics allowlist

[CatherineF-dev]

Cilium Feature Proposal

Thanks for taking time to make a feature proposal for Cilium! If you have usage questions, please try the slack channel and see the FAQ first.

Is your proposed feature related to a problem?
#20810

If so, please describe the problem
Some cilium metrics cardinality are very high. Current open-telemetry collector will read the whole metrics payload, which can cause open-telemetry collector OOM.

Describe the feature you'd like
Add a flag --metrics-allowlist to only expose certain cilium metrics

Include any specific requirements you need

(Optional) Describe your proposed solution

Please complete this section if you have ideas / suggestions on how to implement the feature. We strongly recommend discussing your approach with Cilium committers before spending lots of time implementing a change.

For longer proposals, you are welcome to link to an external doc (e.g. a Google doc). We have a Cilium Feature Proposal template to help you structure your proposal - if you would like to use it, please make a copy and ensure it's publicly visible, and then add the link here.

Once the CFP is close to being finalized, please add it as a PR to the design-cfps repo for final approval.

• https://github.com/cilium/cilium/blob/main/pkg/metrics/metrics.go#L1002

Tasks

Beta Give feedback

No tasks being tracked yet.

[ti-mo]

Hi @CatherineF-dev, thanks for the proposal! Could you elaborate on what exactly --metrics-allowlist's effect would be? I'm not sure if an agent flag is useful for this beyond specifying 2-3 metrics, or this will grow huge. (there are limits on how long arguments can be)

Could you flesh this out a bit more?

[CatherineF-dev]

Explicitly list all exposed metrics in --metrics-allowlist.

For example, cilium_k8s_client_api_latency_time_seconds_bucket causes some OOMs. After updating --metrics-allowlist, this metric will not be exposed. Not only this metric, found other cilium metrics cardinality is high as well.

[CatherineF-dev]

After PR prometheus/client_golang#1327, cilium will be able to use metrics-allowlist.

[github-actions]

This issue has been automatically marked as stale because it has not
had recent activity. It will be closed if no further activity occurs.

[github-actions]

This issue has not seen any activity since it was marked stale.
Closing.