[SOLVED] PVE/vzdump ignores --dumpdir option

[budy]

Can it be, that there has been an error introduced in latest version of PVE? I do have configured a zpool and on that zpool is a folder, which I have set to be used for backups. However, every time I try to run a backup and select the folder, I am seeing this:



As you can see, the amount of available space is the amount of space on the local boot volume, where PVE has been installed on, but not the amount of space available on the zpool, which is currently about 1.48TB. When trying to backup a guest to any of the locations vmBackup or Backups, which both reside on the zpool, the backups are attempted on the boot volume instead and fail.

The same applied to trying to run vzdump manually like this:

vzdump 103  --node vm --dumpdir /vmpool/vmBackup/dump --compress lzo  --mode snapshot

This also creates the backup on the boot volume. The PVE version is:

root@vm:/vmpool# pveversion
pve-manager/6.1-8/806edfe1 (running kernel: 5.3.10-1-pve)

storage.cfg looks like this:

dir: local
        path /var/lib/vz
        content vztmpl,iso,backup

lvmthin: local-lvm
        thinpool data
        vgname pve
        content rootdir,images

zfspool: VMs
        pool vmpool
        content images,rootdir
        mountpoint /vmpool
        sparse 1

dir: vmBackup
        path /vmpool/vmBackup
        content vztmpl,backup,rootdir,snippets
        maxfiles 1
        shared 0

And what's even worse… after a reboot none of the LXCs I had configured were able to start, although their ZFS subvols had been present an accessible.

 

[budy]

So… it turned out, that my pve host crashed due to the mail gateway LXC creating repeatedly OOM-killer invocations. After that reboot the zpool didn't got mounted in time and then the mount point got "occupied" and thus the ZFS containing the folder didn't get mounted.

So… I ended up shutting down all the pve services since I couldn't orderly unmount the zpool. Once I stopped pvedaemon, I was able to properly unmount/mount the zpool. Afterwards, a proper host reboot brought everything back to normal.