GKE Regional Cluster support doesn't work

[apenney]

I ran into a small issue when trying to test out tidb in GKE. I created a regional cluster (in us-east-4) and discovered when I deploy the cluster itself that it constantly tries to create a disk in us-east-4c and then tries to schedule the pod into us-east-4b.

The pv has:

  labels:
    failure-domain.beta.kubernetes.io/region: us-east4
    failure-domain.beta.kubernetes.io/zone: us-east4-c

I'm not a kubernetes expert but it looks like we lose those labels when scheduling pods and it then doesn't know it must schedule this pod in the same location as the storage. Has anyone else seen this or have any ideas what we're doing wrong?

For now I'll delete my cluster, make a zonal cluster, and retest.

[gregwebs]

duplicate of #241

Yes, for now you must use a single zone cluster. We are looking into this and should get to the bottom of the issue next week.

[gregwebs]

This should be fixed by using the charts from master now.