[CRIU] [Announce] Stable 1.3.1 release

Tue Sep 16 12:51:12 PDT 2014

But maybe the easiest way is to simply build my system so this doesn't happen.  I really don't care what host OS and what container template I use.  I just want to be able to suspend and then resume a container, which then can be attached to (lxc-attach).  I don't care which version of LXC or criu I use.  I just need a short list of ingredients.

Thanks,
Krystof

-----Original Message-----
From: Pavel Emelyanov [mailto:xemul at parallels.com] 
Sent: Tuesday, September 16, 2014 12:03 PM
To: Zmudzinski, Krystof C; Tycho Andersen
Cc: criu at openvz.org
Subject: Re: [CRIU] [Announce] Stable 1.3.1 release

On 09/16/2014 02:12 AM, Zmudzinski, Krystof C wrote:
> I'm trying to build a system where I can suspend a container and resume it.  I have tried different things but so far I can only suspend a container using the example on http://criu.org/LXC.
> I can also resume it but it either goes away if I use the -d option with criu restore or it stays there as a child of criu restore without the -d option and lxc-info -n cn says that it is stopped.  This is not optimal since I'd like to use lxc-attach to take a look inside the restored container.
> 
> I used the default LXC package and the 1.1.0-alpha1 with the same results.  Furthermore, lxc-checkpoint in 1.1.0-alpha1 gives me the following errors even when criu dump executed directly succeeds:
> Error (files-reg.c:601): Unaccessible path opened 38:144946, need 
> 2049:2244303

This means, that task opened a file by some name, but then something happened and this name got re-used by some other file.
This "something" can be an overmount. Typically this happens when init opens /dev/null from /, then /dev gets mounted with devtmpfs or tmpfs and new /dev/null is planted on it.

If the issue is _really_ with /dev/null you can workaround this with --evasive-devices option. It will tell criu to open devices under any name that leads to them.

If this is the regular file that got its name "overridden", then we'd need to patch criu.

> Error (cr-dump.c:1603): Dump files (pid: 6560) failed with -1 Error 
> (cr-dump.c:1953): Dumping FAILED.
> 
> I'm willing to try anything but so far, for some reason, I'm not very successful.  
> 
> Krystof
> 
> -----Original Message-----
> From: Tycho Andersen [mailto:tycho.andersen at canonical.com]
> Sent: Monday, September 15, 2014 3:04 PM
> To: Zmudzinski, Krystof C
> Cc: criu at openvz.org; Pavel Emelyanov
> Subject: Re: [CRIU] [Announce] Stable 1.3.1 release
> 
> Hi Krystof,
> 
> On Mon, Sep 15, 2014 at 06:12:28PM +0000, Zmudzinski, Krystof C wrote:
>> The instructions (http://bazaar.launchpad.net/~tycho-s/+junk/snapshot-instructions/view/head:/README) are not complete or assume that my system is configured in certain way.
>>
>> First, I had problems with cgroup, then this:
>>       lxc-start 1410804526.880 ERROR    lxc_apparmor - No such file or directory - failed to change apparmor profile to lxc-container-default-with-mounting
>>       lxc-start 1410804526.880 ERROR    lxc_sync - invalid sequence number 1. expected 4
>>       lxc-start 1410804526.880 ERROR    lxc_start - failed to spawn 'u1'
>>       lxc-start 1410804526.880 ERROR    lxc_cgfs - Device or resource busy - cgroup_rmdir: failed to delete /cgroup/lxc/u1-1
>>       lxc-start 1410804531.885 ERROR    lxc_start_ui - The container failed to start.
>>       lxc-start 1410804531.885 ERROR    lxc_start_ui - To get more details, run the container in foreground mode.
>>       lxc-start 1410804531.885 ERROR    lxc_start_ui - Additional information can be obtained by setting the --logfile and --logpriority options.
>>       lxc-start 1410804731.572 ERROR    lxc_cgfs - Could not find writable mount point for cgroup hierarchy 3 while trying to create cgroup.
>>       lxc-start 1410804731.572 ERROR    lxc_start - failed creating cgroups
>>       lxc-start 1410804731.599 ERROR    lxc_start - failed to spawn 'u1'
>>       lxc-start 1410804736.604 ERROR    lxc_start_ui - The container failed to start.
>>       lxc-start 1410804736.604 ERROR    lxc_start_ui - To get more details, run the container in foreground mode.
>>       lxc-start 1410804736.604 ERROR    lxc_start_ui - Additional information can be obtained by setting the --logfile and --logpriority options.
> 
> Hmm, this seems like lxc (or some dependency) is installed incorrectly. Do you happen to have cgroup-lite installed?
> 
> Tycho
> .
> 