Closed YuichiroMaeyama closed 4 years ago
Not a bug. LXD properly invoked CRIU which then failed to dump your container with a pretty clear error.
Note that even if you get past that network dumping error, there's no way you'll be able to dump a bionic container. Both systemd and AppArmor in those cause structures and sockets that CRIU cannot handle at this time.
thank you for your quick response. I understood that it was a problem on the CRIU side. So I have two questions
Will live migration succeed if I use a distribution other than Bionic Container? What distribution should I use for live migration?
What are your plans and progress in discussing this issue with the CRIU, and do you plan to incorporate live migration support into milestones in the future? I would appreciate it if you could tell me your plan.
We used to have funding to work on CRIU as mostly a research project type thing, that's what got us our current integration and a number of fixes for CRIU but that is no longer the case and we don't have anyone on the team working on this at this time.
Busybox/Alpine are the best for checkpoint/restore at this time, so long as you don't run any recent services on it. On the network side, currently only containers without a network device are working as I believe our modern network device logic doesn't work with it at this time.
So it can be used for a demo of CRIU but that's about the extent of its use at this time. There are active users of this mind you, but they have extremely specific workloads which just happen to tick all the right boxes.
@intrepidsilence until CRIU/liblxc get fixed to handle the current NIC devices, stateless migration is your best bet. You could do a first couple of pass with lxc copy SRC DEST --refresh
to get you as close as possible, then stop the source, do another refresh and start the destination.
@stgraber Thank you so much for the quick reply.
Required information
Issue description
Live migration fails
Steps to reproduce
Step one nodeA:$ lxc list +--------------+---------+---------------------+------+-----------+-----------+----------+ | NAME | STATE | IPV4 | IPV6 | TYPE | SNAPSHOTS | LOCATION | +--------------+---------+---------------------+------+-----------+-----------+----------+ | bionic-nodeA | RUNNING | 10.47.93.199 (eth0) | | CONTAINER | 0 | nodeA | +--------------+---------+---------------------+------+-----------+-----------+----------+
Step two nodeB:$ lxc list +------+-------+------+------+------+-----------+----------+ | NAME | STATE | IPV4 | IPV6 | TYPE | SNAPSHOTS | LOCATION | +------+-------+------+------+------+-----------+----------+
Information to attach
dmesg
)lxc info NAME --show-log
)lxc config show NAME --expanded
)lxc monitor
while reproducing the issue)