Add fsGroupChangePolicy to backup cronjob of toolbox helmchart
Summary
As GitLab backup disk gets bigger, Kubernetes Azure (1.23) complains about creating the backup pod from the CronJob.
As it was done for gitaly here !2786 (merged) I would highly encourage we do it for the toolbox as well. This is also mentioned on Azure troubleshooting page https://learn.microsoft.com/en-us/troubleshoot/azure/azure-kubernetes/fail-to-mount-azure-disk-volume#error5
Steps to reproduce
Not sure how to reproduce, but we have a very big disk and the error we have from kubectl -n GITLAB_NAMESPACE get events is:
MountVolume.SetUp failed for volume "pvc-XXXXXXXXXXXXXXXXXXXXXXXXX" : applyFSGroup failed for vol /subscriptions/XXXXXXXXXXXXXXXXXXXXXX/resourceGroups/XXXXXXXXXXXXXXXXXXXXX/providers/Microsoft.Compute/disks/XXXXXXXXXXXXXXXXXX: readdirent /var/lib/kubelet/pods/XXXXXXXXXXXXXXXXXX/volumes/kubernetes.io~csi/XXXXXXXXXXXXXXXXXX/mount/artifacts/05/ad/05ada863a4cf9660fd8c68e2295f1d35b2264815f5b605003d6625bd9e0492cf/2021_08_27/310161/238537: input/output error
Configuration used
for toolbox, we use the default from this helmchart.
(Paste sanitized configuration here)
Current behavior
backup jobs not working
Expected behavior
backup job should work
Versions
- Platform:
- Cloud: AKS 1.23
- gitlab: 1.15.1
- gitlab helm chart version: 6.5.1
Edited by Everton Seiei Arakaki