Quoting Tamas Papp (tom...@martos.bme.hu): > > On 02/21/2014 05:49 PM, Serge Hallyn wrote: > > Sorry, I'm afraid not :( But so move_pid is what is failing... > > which means the create and chown calls have already succeeded. > > > > If you can tell me that today, with latest uptodate packages from > > ppa on precise, it still fails, then I will - probably on monday - > > try to reproduce and nail it down. > > Hmm, it is working, if cgmanager is stopped: > > Without: > > lxc-attach 1393025778.900 WARN lxc_log - lxc_log_init called > with log already initialized > lxc-attach 1393025778.901 WARN lxc_log - lxc_log_init called > with log already initialized > lxc-attach 1393025778.901 ERROR lxc_cgmanager - call to > cgmanager_move_pid_sync failed: invalid request > lxc-attach 1393025778.901 ERROR lxc_cgmanager - Failed to enter > group /lxc/elf-mail > lxc-attach 1393025778.902 ERROR lxc_attach - error communicating > with child process > > With: > > root@elf:~/1# cat without > lxc-attach 1393025798.197 WARN lxc_log - lxc_log_init called > with log already initialized > lxc-attach 1393025798.264 WARN lxc_attach - could not change > directory to '/root/1' > lxc-attach 1393025798.265 INFO lxc_apparmor - changed apparmor > exec profile to lxc-container-default > > ii lxc > 1.0.0+master~20140220-2000-0 amd64 Linux > Containers userspace tools > ii lxc > 1.0.0+master~20140220-2000-0 amd64 Linux > Containers userspace tools
Ok, thanks Tamas - this is a bug in how the cgproxy starts up on an older kernel. As a workaround until I get it fixed, you can sudo stop cgproxy sudo stop cgmanager sudo rm -rf /sys/fs/cgroup/cgmanager* then in one terminal: sudo /sbin/cgmanager --debug and in another: sudo /sbin/cgproxy --debug -serge _______________________________________________ lxc-users mailing list lxc-users@lists.linuxcontainers.org http://lists.linuxcontainers.org/listinfo/lxc-users
