For all the below scenarios, am seeing that irrespective of giving RWO in the PVC , the same portworx volume is r/w accessible from multiple nodes. Is the kubernetes volume accessMode RWO [r/w once from a single node] in the PVC not relevant in the context of portworx, please comment:
Scenarios:
a) shared: “true” in px storage class [ using the storage class provided by portworx, ‘portworx-shared-sc’ ] & access mode RWO in the pvc used by the deployment.
b) set shared: “false” in px storage class [ using the storage class provided by portworx, ‘portworx-shared-sc’ ] & access mode RWO in the pvc used by the deployment. For this scenario, have seen a slight variation in the behaviour. If I set node affinity to a non-storage node, the 2nd replica if bought up sequentially fails to mount the same volume.
Please comment, is RWO accessMode in the PVC ignored in the context of portworx.