Anyone using Bitbucket DC on Kubernetes that scaled more than 1?

Question

Does anyone have this working where you have more than one Bitbucket server (HA), and how did you deal with helm PersistentVolumeClaim?

The other idea I am thinking of is using one Helm release per Bitbucket instance on Kubernetes to get around this problem. This may even give me the option to upgrade one at a time or take one down when needed rather than using a rolling update that can go wrong and take down all the HA servers.

Any ideas/tips/examples?

Yevhen_5guu4k · Answer

https://github.com/atlassian/data-center-helm-charts/blob/main/src/main/charts/bitbucket/values.yaml#L223 you need to set this to true which will enable volumeClaimTemplates in Bitbucket sts spec.

Mark Hubers · Answer

Or should I even care if the localhome gets deleted at times as I not clear what happens to things like cache data etc.

Mark Hubers · Answer

The shared home is RWX and that one is getting mounted on the 2nd pod.  It is the local home that is RWO, as each bitbucket pod needs its own.

bitbucket-sharedhome-pv 2Gi RWX Retain Bound bitbucket/bitbucket-sharedhome-pvc efs-sc  47h
pvc-c24e8df7-3b7d-4e98-95b6-8c2a4df18b35 20Gi RWO Delete Bound bitbucket/bitbucket-localhome-pvc ebs

I cannot find a way to give each bitbuket pod it own localhome using ebs-csi driver