[lxc-users] New live migration issues

[Tycho Andersen]

On Mon, May 16, 2016 at 08:52:34AM +0000, Jamie Brown wrote:
> Hello again,
> 
> I’m now running LXD 2.0.0 on Ubuntu 16.04 LTS after having some amount of success with a previous BETA on 14.04.
> 
> However, after a fresh install, the first attempt at a live migration has failed;
> 
> lxc launch ubuntu:xenial host2:migrate
> lxc move migrate host3:
> error: Error transferring container data: restore failed:
> (00.499626)      1: Error (mount.c:2406): mnt: Can't mount at ./sys/kernel/debug: Invalid argument
> (00.515123) Error (cr-restore.c:1350): 8965 exited, status=1
> (00.573498) Error (cr-restore.c:2182): Restoring FAILED.
> 
> The end of the migration log on host3;
> (00.499574)      1: mnt: 129:./sys/fs/cgroup/net_cls,net_prio private 1 shared 0 slave 0
> (00.499588)      1: mnt:           Mounting securityfs @./sys/kernel/security (0)
> (00.499593)      1: mnt:           Bind /sys/kernel/security/ to ./sys/kernel/security
> (00.499603)      1: mnt: 276:./sys/kernel/security private 0 shared 0 slave 1
> (00.499610)      1: mnt:           Mounting debugfs @./sys/kernel/debug (0)
> (00.499615)      1: mnt:           Bind /sys/kernel/debug/ to ./sys/kernel/debug
> (00.499626)      1: Error (mount.c:2406): mnt: Can't mount at ./sys/kernel/debug: Invalid argument
> (00.515123) Error (cr-restore.c:1350): 8965 exited, status=1
> (00.555143) Switching to new ns to clean ghosts
> (00.572964) uns: calling exit_usernsd (-1, 1)
> (00.573065) uns: daemon calls 0x4523c0 (8960, -1, 1)
> (00.573108) uns: `- daemon exits w/ 0
> (00.573458) uns: daemon stopped
> (00.573498) Error (cr-restore.c:2182): Restoring FAILED.
> 
> Both hosts have criu installed and are running the same kernel version (4.4.0-22-generic). The CPU and hardware profile is absolutely identical.
> 
> Both hosts are using ZFS file storage.
> 
> Stop and copy between hosts works without issue. The criu dump also completes successfully, so the issue appears to be with criu restore.
> 
> Any ideas?

In addition to the bug Stéphane linked to, I've also posted v3 of the
patchset to fix this (which I hope to be the final version) on the
CRIU mailing list:

https://lists.openvz.org/pipermail/criu/2016-May/028432.html

Tycho

> Thanks,
> 
> Jamie

> _______________________________________________
> lxc-users mailing list
> lxc-users at lists.linuxcontainers.org
> http://lists.linuxcontainers.org/listinfo/lxc-users