Website: Issue with k8s.io/docs/concepts/storage/volumes/

[abitrolly]

This is a...

• [x] Bug Report

Problem:
No info how to inspect volume usage or how to get alert that volume is out of space.

Page to Update:
https://kubernetes.io/docs/concepts/storage/volumes/

[zhangxiaoyu-zidif]

/cc @jingxu97
/sig-storage

[zhangxiaoyu-zidif]

As I know now, only local storage supports storage quota so far.
So the issue you mentioned is a future feature of k8s, I think...

[zhangxiaoyu-zidif]

https://github.com/kubernetes/community/blob/master/contributors/design-proposals/storage/volume-metrics.md

https://github.com/kubernetes/community/blob/master/contributors/design-proposals/storage/local-storage-overview.md

These links may be helpful. =)

[abitrolly]

Thanks. I don't know how much I spent looking on the internet just to discover that this is impossible. :D

Now with that info I don't understand how GKE can be production ready. What happens to Pod if its PVC/PV runs out of space?

[zhangxiaoyu-zidif]

I do not know GKE much, but I know some about Ceph, if we run out of PVC's space, it will continue to use the space until it runs out of Ceph's space. In this case, PVC/PV's disk quota is fake.

[zacharysarah]

@abitrolly 👋 This issue is more of a feature request than a bug—and like @zhangxiaoyu-zidif pointed out, it's already on the roadmap. I'm going to go ahead and close this issue, since it's not actionable (yet) and plans for its actionability are accounted for in roadmap development.

[jingxu97]

@abitrolly We recently added a feature to expose PVC volume's disk usage information through summary API. There are a couple of ways to access this data

1. Run "kubectl proxy" first. Then " curl localhost:8001/api/v1/proxy/nodes/:10250/metrics"
2. Ssh to the node, Run "curl -k http://localhost:10255/metrics"

You will see metrics data like this

# TYPE kubelet_volume_stats_capacity_bytes gauge
kubelet_volume_stats_capacity_bytes{namespace="default",persistentvolumeclaim="data-0"} 1.0501771264e+10

We plan to continue to work on exposing metrics in more convenient ways in 1.9. Please let me know if you have any issues or questions. Thanks!

[abitrolly]

Thanks. I will explore the options as soon as I get back to it. I still use standard dashboard application to troubleshoot the cluster, and I need to figure out how to add this metric here. Checking it by hand is hard.