PV reservation with Local PV ZFS? #3819
Unanswered
fabriziogiudici
asked this question in
General
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hello. I'm working to evaluate OpenEBS as an improved way of managing volumes in a Kubernetes project that presently uses native storage supplied by the cloud provider. My first step is to integrate Local PV ZFS; when and if it will be consolidated I'll try Replicated PV Mayastor.
I followed the instructions and had the system up and running with no big troubles, but I have two issues that so far I wasn't able to fix:
pvc-__UUID__
and I'd like to have a custom name, as it happens in my current system.reclaimPolicy: Retain
), but if I undeploy/redeploy a package that is using a volume (or if I rebuild the system from scratch) new empty persistent volumes are created each time; the old ones are still in the zpool, but unused.This to me is really a problem since I can't resync the volumes from a backup every time I re-create the cluster, as it would take several minutes. Of course in production undeploying / rebuilding the system is something that I seldom do, but in the test environment it might happen a few times in a day when there is intense development.
Below the configuration files I'm using.
This is the StorageClass I've deployed:
This is the PVC:
I've tried to add a
volumeName: portainer
which, according to the specs, should set the created volume, but the pvc just hangs inPending
state.I've tried to configure the thing with the usual approach described at https://kubernetes.io/docs/concepts/storage/persistent-volumes/#reserving-a-persistentvolume:
and having an explicit PV (as documented also at https://kubernetes.io/blog/2019/01/15/container-storage-interface-ga/#pre-provisioned-volumes):
A pv with the desided name appears and it seems bound to the claim:
Also the pvc seems ok:
But the pod can't mount it:
And actually it doesn't appear in the
zfs list
on the node where the zpool is accessible.I've read https://github.com/openebs/openebs/issues/479, but the proposed solution is cumbersome and not easily automatable. I have playbooks capable to create the infrastructure and deploy k8s from scratch without operator intervention and I want them to stay.
Please:
Thanks.
Beta Was this translation helpful? Give feedback.
All reactions