We are deploying to a Kubernetes Environment using RookCeph.

We hit an issue that the Rook Ceph operator was not requesting the block storage to be provisioned. After some digging we found out that the Operator was hitting an out of memory issues. We increased the memory from 128MB to 512MB (1024MB limit). We then deleted the PVCs and Pods for the deployment that failed. The Pods came back up with the PVC requests and everything worked after (a very stressful) ten mins.

Simple fix but it was not clear what the issue was at first.

By Chris Phillips on August 22, 2018.

Canonical link

Exported from Medium on April 6, 2019.