[lxc-users] LXC not responsive after update

[Viktor Trojanovic]

Hi Bostjan,

I sent a reply with an attachment a week ago but it still was not 
approved. To not delay this any further, I just pasted the result of 
below commands here:

strace: http://pastebin.com/FXr4WJqG
lsof: http://pastebin.com/bHHtkCdM

Thanks for the commands, by the way. I read through the result but I 
can't make anything of it.

Viktor

On 22.01.2016 04:27, Bostjan Skufca wrote:
> Do not add -eopen argument, as it limits your view.
>
> Do this:
> - strace -f -tt -r lxc-ls -f
>
> And while strace is running and your program is hanging, find its pid 
> (of lxc-ls, not strace) and run lsof -p PID (if it will be hanging on 
> socket, this will give you a hint what that socket is).
>
> b.
>
>
> On 21 January 2016 at 23:42, Viktor Trojanovic <viktor at troja.ch 
> <mailto:viktor at troja.ch>> wrote:
>
>     Hi Bostjan,
>
>     Here is my strace to "lxc-ls -f". It looks the same with the other
>     commands, at least the last line is identical.
>
>     # strace -eopen lxc-ls -f
>     open("/etc/ld.so.cache", O_RDONLY|O_CLOEXEC) = 3
>     open("/usr/lib/libpython3.5m.so.1.0", O_RDONLY|O_CLOEXEC) = 3
>     [...]
>     open("/usr/share/lxc/config/archlinux.common.conf", O_RDONLY) = 6
>     open("/usr/share/lxc/config/common.conf", O_RDONLY) = 7
>     open("/usr/share/lxc/config/common.conf.d/",
>     O_RDONLY|O_NONBLOCK|O_DIRECTORY|O_CLOEXEC) = 8
>
>     I can't make much of that. I see that it cannot find a few files but
>     in my understanding these are optional anyway.
>
>     Maybe you see something I don't?
>
>     Viktor
>
>     On Thu, Jan 21, 2016 at 6:33 PM, Bostjan Skufca <bostjan at a2o.si>
>     wrote:
>     > strace command is your friend.
>     >
>     > b.
>     >
>     >
>     > On 21 January 2016 at 12:12, Viktor Trojanovic <viktor at troja.ch>
>     wrote:
>     >>
>     >>
>     >>
>     >> On 20.01.2016 23:50, Fajar A. Nugraha wrote:
>     >>
>     >> On Thu, Jan 21, 2016 at 5:49 AM, Fajar A. Nugraha
>     <list at fajar.net <mailto:list at fajar.net>> wrote:
>     >>>
>     >>> On Thu, Jan 21, 2016 at 5:23 AM, Viktor Trojanovic
>     <viktor at troja.ch <mailto:viktor at troja.ch>>
>     >>> wrote:
>     >>>>
>     >>>> I just did a system upgrade on my Arch System which included
>     updating
>     >>>> the kernel, systemd and lxc to the newest versions. After
>     having done so, I
>     >>>> cannot interact with my Linux container any longer. The
>     system within the
>     >>>> container still seems to work fine and can be contacted from
>     outside (Samba
>     >>>> server) but if I try to use one of the lxc commands to query
>     or otherwise
>     >>>> interact with the container (e.g. lxc-ls -f, lxc-stop,
>     lxc-console,
>     >>>> lxc-attach), the command hangs until I cancel it with CTRL+C.
>     >>>>
>     >>>> I get the following message when cancelling lxc-ls -f (as root):
>     >>>>
>     >>>> ^CTraceback (most recent call list):
>     >>>>   File "/usr/bin/lxc-ls", line 432, in <module>
>     >>>>     containers = get_containers(root=True)
>     >>>>   File "user/bin/lxc-ls", line 261, in get_containers
>     >>>>     if container.controllable:
>     >>>> KeyboardInterrupt
>     >>>>
>     >>>> Regular lxc-ls works normal, by the way.
>     >>>>
>     >>>> I can probably just reboot the server but I still wanted to
>     ask around
>     >>>> if anyone has an idea why this is happening and what I could
>     do except
>     >>>> rebooting to regain control of LXC? I tried already systemctl
>     restart lxc
>     >>>> but that doesn't help.
>     >>>>
>     >>>
>     >>>
>     >>> Do you have lxcfs installed? If yes, this should be a know
>     issue. When
>     >>> you restart lxcfs, all existing running containers that use it
>     will be
>     >>> unable to access lxcfs-provided resources. AFAIK restarting
>     lxc service does
>     >>> restart running containers.
>     >>>
>     >>
>     >> That should be: AFAIK restarting lxc service does NOT restart
>     running
>     >> containers.
>     >>
>     >>
>     >>> Try killing one of those containers (lxc-stop -k -n ...),
>     start it, and
>     >>> then test again. If it works, do the same for other containers.
>     >>>
>     >>> --
>     >>> Fajar
>     >>
>     >>
>     >> Hi Fajar,
>     >>
>     >> I'm not using lxcfs, just regular directory based containers.
>     >>
>     >> I cannot do anything with the containers.
>     >>
>     >> lxc-stop -k
>     >> lxc-ls -f
>     >> lxc-attach
>     >> lxc-console
>     >> systemctl stop lxc (or restart lxc)
>     >>
>     >> All these commands hang forever, I have to interrupt with
>     CTRL+C and no
>     >> output is written to the logs. But the container system keeps
>     working fine,
>     >> so does the host system.
>     >>
>     >> Viktor
>     >>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linuxcontainers.org/pipermail/lxc-users/attachments/20160127/ac750633/attachment.html>