Bug 2162333
| Summary: | PVC created using non default storage class on fresh cluster | ||||||
|---|---|---|---|---|---|---|---|
| Product: | Container Native Virtualization (CNV) | Reporter: | Roni Kishner <rkishner> | ||||
| Component: | Storage | Assignee: | Adam Litke <alitke> | ||||
| Status: | CLOSED ERRATA | QA Contact: | Yan Du <yadu> | ||||
| Severity: | high | Docs Contact: | |||||
| Priority: | unspecified | ||||||
| Version: | 4.13.0 | CC: | dollierp, gouyang, jpeimer, yadu | ||||
| Target Milestone: | --- | ||||||
| Target Release: | 4.13.0 | ||||||
| Hardware: | All | ||||||
| OS: | All | ||||||
| Whiteboard: | |||||||
| Fixed In Version: | Doc Type: | If docs needed, set a value | |||||
| Doc Text: | Story Points: | --- | |||||
| Clone Of: | Environment: | ||||||
| Last Closed: | 2023-05-18 02:56:40 UTC | Type: | Bug | ||||
| Regression: | --- | Mount Type: | --- | ||||
| Documentation: | --- | CRM: | |||||
| Verified Versions: | Category: | --- | |||||
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |||||
| Cloudforms Team: | --- | Target Upstream Version: | |||||
| Embargoed: | |||||||
| Attachments: |
|
||||||
|
Description
Roni Kishner
2023-01-19 11:28:27 UTC
Hi, Denis Could you please help to check whether the installation deploy script changed in 4.13? Hi, AFAIK, our installation script didn't change in 4.13. Created attachment 1943261 [details]
storage classes definitions
This looks like a deeper problem then a new cluster. I just created a new VM from a template without defining the storage class and it used "standard-csi" instead of the default storage class which is "hostpath-csi-basic".
attaching a list of both storage classes definition
(In reply to Roni Kishner from comment #3) > Created attachment 1943261 [details] > storage classes definitions > > This looks like a deeper problem then a new cluster. I just created a new VM > from a template without defining the storage class and it used > "standard-csi" instead of the default storage class which is > "hostpath-csi-basic". > attaching a list of both storage classes definition Ignore the comment, it was my mistake, we create a new DV/PVC from dataSource but the template still looks at the original PVC created so it used a wrong storageclass TL DR: It's because of the cluster setup flow (maybe we need to unset 'default' of 'standard-csi' earlier in the flow) From my findings of 4.13 deployment flow: Once we deployed OCP, but not yet deployed CNV, 'standard-csi' is a default storage: [cloud-user@ocp-psi-executor ~]$ oc get sc NAME PROVISIONER RECLAIMPOLICY VOLUMEBINDINGMODE ALLOWVOLUMEEXPANSION AGE csi-manila-ceph manila.csi.openstack.org Delete Immediate false 60m standard-csi (default) cinder.csi.openstack.org Delete WaitForFirstConsumer true 61m Then we deploy CNV and data import cron DVs start to import: [cloud-user@ocp-psi-executor ~]$ oc get pvc -n openshift-virtualization-os-images NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE centos-stream8-2f16c067b974 Bound pvc-52e914ea-5d25-4fc7-8b3e-678bd01dc171 30Gi RWO standard-csi 84s centos-stream9-23d7d288eb34 Bound pvc-661ed6ba-6915-4d23-a6d2-d76e58d7991c 30Gi RWO standard-csi 86s centos7-680e9b4e0fba Bound pvc-6caab12b-621c-428c-b228-cdb861021d43 30Gi RWO standard-csi 85s fedora-56ccabc01cbe Bound pvc-8ea4e2b4-2fa3-48ed-b130-b416856aa7ba 30Gi RWO standard-csi 81s rhel8-e004563ff8cf Bound pvc-9ead3860-e602-40fa-937b-ee18f3dd6496 30Gi RWO standard-csi 87s rhel9-2a70e7a1fadc Bound pvc-1a415f8f-5119-4a82-ae9f-1fe6a09888b0 30Gi RWO standard-csi 87s And that exact moment the storage classes are: [cloud-user@ocp-psi-executor ~]$ oc get sc -w NAME PROVISIONER RECLAIMPOLICY VOLUMEBINDINGMODE ALLOWVOLUMEEXPANSION AGE csi-manila-ceph manila.csi.openstack.org Delete Immediate false 89m local-block-hpp kubernetes.io/no-provisioner Delete WaitForFirstConsumer false 21m local-block-ocs kubernetes.io/no-provisioner Delete WaitForFirstConsumer false 21m ocs-storagecluster-ceph-rbd openshift-storage.rbd.csi.ceph.com Delete Immediate true 11m ocs-storagecluster-cephfs openshift-storage.cephfs.csi.ceph.com Delete Immediate true 11m standard-csi (default) cinder.csi.openstack.org Delete WaitForFirstConsumer true 90m 'standard-csi' is a default, 'hostpath-csi-basic' is not yet deployed, so data import cron DVs were imported as expected and used the default storage class. In the fresh 4.12 cluster, it also picked the 'standard-csi' storage class: OCP-4.12.3 + CNV-v4.12.1-39 [cloud-user@ocp-psi-executor ~]$ oc get pvc -n openshift-virtualization-os-images NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE centos-stream8-2f16c067b974 Bound pvc-0ed566c0-d997-44e8-b89f-16bd9203a29a 30Gi RWO standard-csi 17m centos-stream9-23d7d288eb34 Bound pvc-1feca034-833c-4586-b711-d56a8d6d61c6 30Gi RWO standard-csi 17m centos7-680e9b4e0fba Bound pvc-0169663c-2d98-479b-98af-7b68de90c1db 30Gi RWO standard-csi 17m fedora-56ccabc01cbe Bound pvc-6c7b51c3-f939-4b30-876a-6c689d237a25 30Gi RWO standard-csi 17m rhel8-e004563ff8cf Bound pvc-22d6ebb7-c023-48ed-b0c8-f59cb99901e0 30Gi RWO standard-csi 17m rhel9-2a70e7a1fadc Bound pvc-ed6dd791-1500-4a60-a05f-c43c16e1e33c 30Gi RWO standard-csi 17m In our pervious(old) cluster, I can see it is using hostpath-csi-basic as it is the default storage class. seems now all the clusters are picking up 'standard-csi' storage class. $ oc get pvc -n openshift-virtualization-os-images NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE centos-stream8-2f16c067b974 Bound pvc-cf184104-840b-498f-98b1-3c5e7ff8ee5e 149Gi RWO hostpath-csi-basic 3d15h centos-stream9-23d7d288eb34 Bound pvc-847f3fd6-be90-4905-a026-c860d75abc29 149Gi RWO hostpath-csi-basic 3d15h centos7-680e9b4e0fba Bound pvc-880795ce-a452-4dc9-a118-4ee0e0b0ccd1 149Gi RWO hostpath-csi-basic 3d15h fedora-56ccabc01cbe Bound pvc-a48e1355-6cc4-4ef0-a0a7-b99670bb26ad 149Gi RWO hostpath-csi-basic 3d15h rhel8-e004563ff8cf Bound pvc-f9854fc3-1049-4c4d-a2d0-c8c27770c57a 149Gi RWO hostpath-csi-basic 3d15h rhel9-2a70e7a1fadc Bound pvc-9c0e0425-d279-4f4d-aaad-e88265908ef7 149Gi RWO hostpath-csi-basic 3d15h Denis, Could we unset the standard-csi as default storage class during deployment? or maybe we can adjust the deployment flow? Could we set the ocs-storagecluster-ceph-rbd (which CNV suggests to use) as the default storage class? Test on CNV v4.13.0.rhel9-1623 $ oc get pvc -n openshift-virtualization-os-images NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE centos-stream8-2f16c067b974 Bound pvc-03f9e46c-533a-4227-a739-f9e01e3b4b37 149Gi RWO hostpath-csi-basic 17m centos-stream9-4e7ee7558c01 Bound pvc-cbce5d78-9d69-4998-b7cf-d92c3613bc41 149Gi RWO hostpath-csi-basic 17m centos7-680e9b4e0fba Bound pvc-a8536fef-6bf7-4c37-96ee-bf4cdb79689a 149Gi RWO hostpath-csi-basic 17m fedora-56ccabc01cbe Bound pvc-aa7bfcee-0187-4125-974b-2d5c148d3275 149Gi RWO hostpath-csi-basic 17m rhel8-b0bacbb4e426 Bound pvc-3a62a6da-e011-4f9c-a66b-11923f489af8 149Gi RWO hostpath-csi-basic 17m rhel9-2a70e7a1fadc Bound pvc-6432008e-62b7-456d-b3d8-1dbae51279e8 149Gi RWO hostpath-csi-basic 17m Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (Moderate: OpenShift Virtualization 4.13.0 Images security, bug fix, and enhancement update), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHSA-2023:3205 |