Situation's a bit complicated so I hope I can make sense here.
I have an OpenShift cluster installed on vSphere (managers and workers). It has Elastic configured to gather logs. Elastic has a Persistent Volume attached, and OpenShift is configured to use vSphere and create a vmdk for this PV and connect it to the worker VM.
Storage-wise, everything is in the same Datastore. Each worker and manager has its own folder (nothing special there) and the shared disk for Elastic has its own folder in the same Datastore (folder was manually created for this). OpenShift is configured to create the vmdk there.
This works until a worker decides to vMotion, causing vSphere to move the vmdk file to the VM's folder on the Datastore and update the disk's path. Then OpenShift creates a new vmdk for the PV in Elastic's folder and connects it to the VM as a new disk and now I have about 10 different vmdks in different folders and VMs.
Also, it turns out that OpenShift does not support Storage vMotion so it seems that if I want to solve this it's going to be some sort of workaround. I thought of setting affinity rules for the VMs but I'm not too happy with this idea.
Is there any elegant way to configure vSphere to keep the disk's path?