Hello, Unused volume is on the new proxmox node. And this Unused logical volumes are empty, livemigration of the machine leads to immediate kernel panic and fsck (after disconnecting LV on second node) is not able to find any filesystem.
What do you mean by empty? Normally, when linstor creates an additional resource, it should initiate a sync from the original resource to the destination resource. You cannot live migrate the resource until the sync is completed and both resources must show up as UpToDate. Did your check the status of the resource on both nodes with drbdtop ? Are you referring to a kernel panic on the guest VM o/s or on the Proxmox host ? Here is interesting, that the allocated space isn't the same on both sides. I think that a side effect of how ThinLVM works. Meaning , it's probably a case that you have written/modified/deleted data over the time on the original resource, so ThinLVM will increase space allocation on the LV (unless you have enabled discard option on pve), where on the new volume it will show only the actual used data as of now. You have to continuously monitor ThinLVM allocated space, just for this very reason. You can easily run onto serious problems, but this is not a DRBD or a LINSTOR issue anyways. the resource files are shared here - https://pastebin.com/V7mjSCar the status output is drbdsetup status vm-103-disk-1 first node vm-103-disk-1 role:Primary disk:UpToDate pve-virt2 role:Secondary peer-disk:UpToDate second node vm-103-disk-1 role:Secondary disk:UpToDate pve-virt1 role:Primary peer-disk:UpToDate Both resource files and the status output look good... and the last info, I tried to resync the volume and in dmesg I have this, please note, that the amount of data to resync after --discard-my-data is far smaller than the volume itself discard-my-data does not initiate a full sync if that's what you mean. It's rather used in situations where split-brain has occurred, so you discard the last changes on one side in favour of the other. If your intention is to initiate a full sync, then consider using "drbdadm invalidate" or recreate meta-data on the 2nd resource by using "drbdadm create-md --force". The proxmox nodes are connected and live migration works, but only for the new VMs, not the older ones. That's an indication that you're probably doing something wrong when you are trying to manually increase the replicas for the old VMs. Might be easier for you to backup old VMs and restore them as new VMs via Proxmox GUI ? Gianni
_______________________________________________ drbd-user mailing list drbd-user@lists.linbit.com http://lists.linbit.com/mailman/listinfo/drbd-user