Add localPV workload dashboard

[rajaSahil]

Signed-off-by: Sahil Raja [email protected]

What this PR does?

• Add dashboard to monitor localPV workload stats.
• Will add service and servicemonitor in the next PR.

NOTE

• These dashboards are built with the help of grafana variables.

To map persistentvolume_claim with localPV device using join, use this expression

• kube_persistentvolumeclaim_info * on(volumename) group_left openebs_size_of_volume

To map workload with localPV device using join, use this expression

• openebs_size_of_volume * on(volumename) group_right(device) kube_persistentvolumeclaim_info * on(persistentvolumeclaim) group_left kube_pod_spec_volumes_persistentvolumeclaims_info * on(pod) group_left namespace_workload_pod:kube_pod_owner:relabel

How dashboards mapping are done?

• To map workload to mountPoint for filesystem dashboards:

  • Metrics hierarchy
    

• To map workload to localPV device for disk dashboards:

  • Metrics hierarchy
    

Variable dependencies:

Screenshots

Note

• This PR is dependent on lvm-metrics. PR: feat(metrics): Expose LVM metrics lvm-localpv#94

Ref

• expose pvc/pod level metrics (if not already exposed by kubelet and cadvisor) lvm-localpv#78

[iyashu]

@kmova @rajaSahil Is it possible to directly import the dashboard json in grafana without doing helm install? I was trying to add above dashboard in my local cluster, but there seems to be some helm template vars being used in dashboard json as well.

[rajaSahil]

@iyashu There are no helm template vars used in the json. However, there were few links used that might have caused the error. I have removed those links. So can you please try again to import the updated json in grafana?

[avishnu]

lgtm

[Ab-hishek]

lgtm