lxc start error on 4.1

[bkraegelin]

Strange behaviour on starting lxc containers on VE 4.1...

I created a test VM with PVE 4.1 and installed some lxc containers (converted from openvz backup). It looks very promising so far.

On reboot of the host (here, HOST is my PVE 4.1 qemu VM running on PVE 2.3) all containers marked as "start at boot" start and run correctly. I have a problem on rebooting/restarting of containers.

Rebooting/restarting a container gives an error:
"Error mounting /dev/loop2 at /media/root/... `mount ...`exited with non-zero exit status 32: mount: wrong fs type, bad option, bad superblock ..."

dmesg shows error:
"EXT4-fs warning (device loop2): ext4_multi_mount_protect:318: MMP interval 42 higher than expected, please wait."
"EXT4-fs warning (device loop2): ext4_multi_mount_protect:331: Device is already on another node."
"EXT4-fs warning (device loop2): ext4_multi_mount_protect:331: MMP failure infor: last update time: 1450254174, last update node: pve, last update device: loop1"

Manually umounting "/media/root/xxxxxxxx-xxxx-xxxx-xxxxxxxxxxxx/" allows starting container directly after it.

Question:
Is there a problem umounting the container on container stop?
Is this a bug?

Additional info:
I have installed Gnome environment ("apt-get install lxde iceweasel") to manage by browser on host console, so perhaps there may be a conflict automounting uor unmounting the image.

Thanks for help
Birger

 

[wbumiller]

Apparently your desktop environment is mounting the disks automatically when they're activated. We enable multiple-mount protection as a safety mechanism when using network storage. You'll have to disable automatic mounting of these devices in your desktop environment's configuration.

 

[bkraegelin]

Thanks for the hint.

I created a new unpriviledged user to login on console and added this user to group "sudo". Now everything works as expected.
(Using unpriviledged user is better anyway.)
Disabled automount in Filemanager->Preferences->Volume Management.