[Lxc-users] kramic on maverick

Trent W. Buck trentbuck at gmail.com
Tue Nov 30 00:49:26 UTC 2010


"Serge E. Hallyn" <serge.hallyn at canonical.com>
writes:

> Quoting Trent W. Buck (twb at cybersource.com.au):
>> I encountered, isolated, and worked around this issue.
>> 
>> Symptom: gettys don't start in the container.  runlevel(8) reports
>> "unknown" instead of "N 2".
>> 
>> Problem: ifupdown 0.6.8ubuntu29.1 introduces this change:
>> 
>>     ifupdown (0.6.8ubuntu29.1) lucid-proposed; urgency=low
>> 
>>       * debian/ifupdown.network-interface.conf: Bring up lo manually, so
>>         that it
>>         comes up even if /etc/network/interfaces is broken (LP: #512253).
>>       * debian/ifupdown.upstart.if-up: Don't emit a duplicate net-device-up
>>         event for lo here, as network-interface.conf will have taken care of
>>         it.
>> 
>> For some reason, upstart DOES NOT generate the event, so when this
>
> Purely guessing, but I assume that normally the kernel creates that
> uevent when lo gets set up?  But with the container, lo is created
> (and uevent sent) before the container's upstart is running?

I have no idea.  Any diagnostics commands you'd like me to run?

> It's important to know where the event is supposed to come from,
> because we're hoping to have the same set of upstart scripts work
> both in a container and on a native host.  So we'll want to find
> a general solution to this - even if it has to be some lxc.conf
> which only runs if it is in a container, and fakes certain events
> it knows won't otherwise happen.

I'm happy to help with that, but my top priority is "make it work at all
in lucid" not "make it work properly in ubuntu+1" :-)





More information about the lxc-users mailing list