Pranshu Srivastava
Pranshu Srivastava
@mrueg Can you please `/lgtm` (if all looks good here)?
Technically, there could be three metrics with the same `kube_pod_container_resource_.+`, and it seems like [this behavior was intentional](https://github.com/kubernetes/kube-state-metrics/blob/2c3ea01cb2c2ca3fa6ea9958d899f497ef14f5c0/internal/store/pod.go#L274-L276), due to the [varying types of resource names](https://github.com/kubernetes/kube-state-metrics/blob/2c3ea01cb2c2ca3fa6ea9958d899f497ef14f5c0/internal/store/pod.go#L252-L269). AFA the transient state...
+1, this should be the expected behavior.
/unassign /remove-lifecycle stale
As @yangjunmyfm192085 pointed out, `metrics-server` simply [lists the nodes, and gets their usage metrics](https://github.com/kubernetes-sigs/metrics-server/blob/d5bca25c88365796fb041470634cb9b31cc19a22/pkg/api/node.go#L84), however, suppressing transient errors could lead to false positives in other cases.
Closing for now. /close
Not an issue with `metrics-server`, closing. /close
/remove-lifecycle rotten /assign @yangjunmyfm192085 Do you have any specific functions in mind that can benefit from this?
@yangjunmyfm192085 So looking at #1009, I understand that this was due to a corner case not being covered in the endless loop inside `decodeBatch`. The only remotely-similar occurence to this...