CRIU crash on LXD live migration and snapshots - Error (criu/sk-netlink.c:73): The socket has data to read

[Robert Roos]

While making a stateful snaphot or doing a live migration CRIU randomly stops working and produces the following errors:

error: snapshot dump failed
(00.200858) Error (criu/sk-netlink.c:73): The socket has data to read
(00.200875) Error (criu/cr-dump.c:1313): Dump files (pid: 2276) failed with -1
(00.224836) Error (criu/cr-dump.c:1628): Dumping FAILED.

Its seems to work better if I first create a normal snaphot and do a stateful snapshot or live migration afterwards. So I did manage to do a live migration and a stateful snaphot on the same system but it appears to be unstable.

Kernel version:

Ubuntu 16.04 - 4.9.0-040900-generic

CRIU version:

2.6-1ubuntu1~ubuntu16.04.2

Guest OS:
Ubuntu 16.04 image

[Kenneth Cheng (chengkunck)]

Suffering from same problem, using Ubuntu 16.04 x86_64 with Linux kernel 4.8.0-30-generic.

CRIU version 2.6

lxd --version: 2.0.8

lxc --version: 2.0.8

Guest container runs Ubuntu 16.04 image from ubuntu remote image repo.

Never succeeded creating a stateful snapshot so far.

[Tycho Andersen (tycho-s)]

Hi guys,

I've marked this as a dupe, since we are already tracking this. It's worth noting that Canonical has shelved development on live migration until we have paying customers for it, so it's unlikely I'll fix this. If you'd like it fixed (and I'd like to fix it :), please contact me or someone at Canonical to buy some support.

Cheers