Some metrics are missing from the kubeletstats receiver

[chenlujjj]

Component(s)

receiver/kubeletstats

Describe the issue you're reporting

Hi Team,

We are using the kubeletstats receiver to collect metrics of pods and containers. But we found that some metrics are missing. For example, container.memory.usage is collected but container.memory.page_faults is missing.

The collector version is 0.79.0 and we didn't exclude any metrics in filter processor.

The config for kubeletstats receiver is:

kubeletstats:
        auth_type: serviceAccount
        collection_interval: 10s
        endpoint: ${K8S_NODE_IP}:10250
        extra_metadata_labels:
        - container.id
        metric_groups:
        - container
        - pod
        - node
        - volume

We have also confirmed that the response of /stats/summary endpoint from kubelet do include container memory page faults infos using following command:

kubectl get --raw /api/v1/nodes/<node_name>:10250/proxy/stats/summary

Any suggestions would be much appreciated.

[github-actions]

Pinging code owners:

• receiver/kubeletstats: @dmitryax @TylerHelmuth

See Adding Labels via Comments if you do not have permissions to add labels yourself.

[TylerHelmuth]

@chenlujjj this is a strange one, I am not able to reproduce the issue. Are you missing any other container.memory.* metrics? The receiver will ignore datapoints if the metric is nil:

opentelemetry-collector-contrib/receiver/kubeletstatsreceiver/internal/kubelet/utils.go

Lines 12 to 17 in 7f452e5

	func recordIntDataPoint(mb *metadata.MetricsBuilder, recordDataPoint metadata.RecordIntDataPointFunc, value *uint64, currentTime pcommon.Timestamp) {
	if value == nil {
	return
	}
	recordDataPoint(mb, currentTime, int64(*value))
	}

, but you're saying the stats/summary definitely has it?

[chenlujjj]

sorry, the problem is probably cause by signalfx exporter we use. i'll close this. Thanks !