Jed Lejosne
Jed Lejosne
> @jean-edouard do you still think we should create an issue for this [#8481 (comment)](https://github.com/kubevirt/kubevirt/pull/8481#issuecomment-1253752850) and experiment with udev rules? Done! Sorry for the delay: https://github.com/kubevirt/kubevirt/issues/8611
Looks like the hotplug-disk pod is not restarted when the VMI is restarted... In that case, this fix won't work.
> Looks like the hotplug-disk pod is not restarted when the VMI is restarted... In that case, this fix won't work. It's actually more subtle than that. When a **stopped**...
> The next best thing would be changing the workflow above to: create virt-launcher pod, hand off to virt-handler, then take care of hotplug-related pods. That was just a matter...
> So the hotplug code you are removing was added in #6079 and the PRs purpose was to allow a VM restart with hotplugged volumes that remain hotplugged after restart....
> Okay so the attachment pod gets re-created, but because in the flow the attachment pod is created before the virt-launcher, we cannot take the selinux labels generated for the...
> Okay I must be confused then, I thought we set the selinux labels in the securityContext in the pod spec, and I was thinking we were reading the value...