Let a pod uses a NFS PV via a PVC. When the pod and the PVC is deleted at the same time, the pod is being killed slowly, while a recycler pod for the NFS PV is started. At some point, both the recycler and the pod may access the same NFS share and mess with each other. The recycler should wait until all pods that use the same PV are deleted. Version-Release number of selected component (if applicable): openshift v3.3.1.3 kubernetes v1.3.0+52492b4 etcd 2.3.0+git How reproducible: ~50% Steps to Reproduce: 1. create 10 separate NFS shares, 10 PVs and 10 claims 2. create 10 pods, each using one claim. The pods randomly create files on the its PVs. 3. delete the pods and claims. Actual results: Recycler is started while the pods are still running and creating new directories. Expected results: Recyler is started *after* appropriate pod that uses the PV that is being recycled is terminated.
I'm going to close this BZ WONTFIX. It's just impossible on NFS. With the merge of 1395271 it should greatly reduce the race risk and risk of old data sticking around too long.