[Lxc-users] [Bulk] Re: Can't connect to an LXC container after restart

[David Parks]

Just about the same time that I received your debug suggestions we had a
datacenter power failure, figures, the container starts fine now. If/when I
can reproduce this issue I'll post all these details. I hadn't tried all the
commands you suggested here.

Thanks!
David


-----Original Message-----
From: Serge Hallyn [mailto:serge.hallyn at ubuntu.com] 
Sent: Tuesday, May 07, 2013 7:19 AM
To: David Parks
Cc: lxc-users at lists.sourceforge.net; 'Ho Duc Ha'; Quy
Subject: [Bulk] Re: [Lxc-users] Can't connect to an LXC container after
restart

Quoting David Parks (davidparks21 at yahoo.com):
> I was playing with a container today, we took the following actions:
> 
> .         Change from static to DHCP IP

Exactly how?  What is the guest distro+release?

> .         Shut down container, add an fstab entry to LXCs fstab file

What entry did you add?

> .         Started, failed to start
> 
> .         Undid all changes, re-configured static IP and removed the fstab
> entry
> 
>  
> 
> Now when I try to start the container I see this in the debug log -o 
> debugfile -l DEBUG:
> 
>  
> 
>      lxc-start 1367826547.900 DEBUG    lxc_conf - umounted
> '/lxc_putold/sys/kernel/security'
> 
>       lxc-start 1367826547.900 DEBUG    lxc_conf - umounted
> '/lxc_putold/proc'
> 
>       lxc-start 1367826547.900 DEBUG    lxc_conf - umounted
> '/lxc_putold/mnt/hdfs'
> 
>       lxc-start 1367826547.900 DEBUG    lxc_conf - umounted
> '/lxc_putold/dev'
> 
>       lxc-start 1367826547.900 DEBUG    lxc_conf - umounted
> '/lxc_putold/run'
> 
>       lxc-start 1367826547.900 DEBUG    lxc_conf - umounted
> '/lxc_putold/sys/fs/cgroup'
> 
>       lxc-start 1367826547.900 DEBUG    lxc_conf - umounted
> '/lxc_putold/sys'
> 
>       lxc-start 1367826547.900 DEBUG    lxc_conf - umounted '/lxc_putold'
> 
>       lxc-start 1367826547.900 INFO     lxc_conf - created new pts
instance
> 
>       lxc-start 1367826547.900 INFO     lxc_conf - set personality to
'0x0'
> 
>       lxc-start 1367826547.900 DEBUG    lxc_conf - drop capability
> 'sys_module' (16)
> 
>       lxc-start 1367826547.900 DEBUG    lxc_conf - drop capability
> 'mac_admin' (33)
> 
>       lxc-start 1367826547.900 DEBUG    lxc_conf - capabilities has been
> setup
> 
>       lxc-start 1367826547.900 NOTICE   lxc_conf - 'testsandboxa01' is
> setup.
> 
>       lxc-start 1367826547.900 INFO     lxc_start - changed apparmor
profile
> to unconfined
> 
>       lxc-start 1367826547.900 NOTICE   lxc_start - exec'ing '/sbin/init'
> 
>       lxc-start 1367826547.917 NOTICE   lxc_start - '/sbin/init' started
> with pid '32017'
> 
>       lxc-start 1367826547.917 WARN     lxc_console - console input
disabled
> 
>       lxc-start 1367826547.917 WARN     lxc_start - invalid pid for
SIGCHLD
> 
>       lxc-start 1367826561.715 DEBUG    lxc_commands - peer has
disconnected
> 
>  
> 
> I can connect to the console, but I get nothing back from the console, 
> I can't SSH to the container, it just appears hung. We had this 
> container running fine for weeks now, only on the restart did I get 
> this, and I'm reasonably confident that we un-did any changes we had made
in our testing.

What does 'sudo lxc-ps -n testsandboxa01' show?

-serge