BCDevOps / platform-services

Collection of platform related tools and configurations
Apache License 2.0
13 stars 29 forks source link

netapp-file-standard not reporting utillization to kubernetes. #722

Closed stewartshea closed 3 years ago

stewartshea commented 4 years ago

Is your feature request related to a problem? Please describe. netapp-file-standard PVCs do not report utilization in the kube_volume_* promtheus metrics endpoints. This makes it impossible for teams to alert based on PVC utilization. This was observed in openshift 3.11. We haven't yet checked OCP4.

Definition of Done PVC's using this storage class are reported in the kubelet metrics and accessible via prometheus & sysdig.

stewartshea commented 4 years ago

Ran a test in our new OCP4 lab cluster and all netapp file based storage classes are working properly. At this point, since OCP3.11 will be decommissioned, I'm not sure it's an issue that is worth fixing on these clusters.

$ oc get pvc  --all-namespaces | grep netapp-file  | grep shea
shea-test                  test                                         Bound     pvc-b8ffee3f-bb14-497c-b39b-99590e2ee141   1Gi        RWX            netapp-file-standard    4m50s

PromQL Query: kubelet_volume_stats_available_bytes{namespace="shea-test"} Result: Element: kubelet_volume_stats_available_bytes{endpoint="https-metrics",instance="hostip:10250",job="kubelet",metrics_path="/metrics",namespace="shea-test",node="nodename",persistentvolumeclaim="test",service="kubelet"} Value: 1073741824

@sbarre-esit thoughts?

StevenBarre commented 4 years ago

Agreed, not worth fixing.

matthieu-foucault commented 3 years ago

kubelet_volume_stats_available_bytes{namespace="0fad32-dev"} does not seem to return any result on sysdig, when looking either in 0fad32-team-persistent-storage or 0fad32-team. Same issue for our other namespaces.

@sbarre-esit @stewartshea are there more steps required on your side for netapp-file-standard and netapp-block-standard usage metrics to be available in sysdig, or am I the one missing some steps?

cc @pbastia @dleard @Maralsotoudehnia @Dianadec

stewartshea commented 3 years ago

@matthieu-foucault I'll have to dig deeper into this, we were focused on getting the team operator deployed but didn't dig deeper into the file / block reporting differences. I'll let you know what we find.

ShellyXueHan commented 3 years ago

@matthieu-foucault you should be able to see PVC metrics from [license-plate]-team-persistent-storage team dashboard now. Please double check and update, thanks!

matthieu-foucault commented 3 years ago

@ShellyXueHan I can confirm that it works, thanks! I tried different namespaces and sysdig teams and I can see usage for both file and block storage.