Hi,

On Sat, Feb 27, 2016 at 11:34:52PM +0100, Ivan Sergio Borgonovo wrote:
> On 02/27/2016 09:37 PM, Evgeni Golov wrote:
> >Hi Ivan,
> 
> >On Sun, Feb 07, 2016 at 03:06:54AM +0100, Ivan Sergio Borgonovo wrote:
> >>got this after upgrade 1:1.0.8-1 -> 1:1.1.5-1
> 
> >>       lxc-start 1454809870.985 ERROR    lxc_apparmor -
> >>lsm/apparmor.c:apparmor_process_label_set:169 - If you really want to start
> >>this container, set
> >>       lxc-start 1454809870.985 ERROR    lxc_apparmor -
> >>lsm/apparmor.c:apparmor_process_label_set:170 - lxc.aa_allow_incomplete = 1
> >>       lxc-start 1454809870.985 ERROR    lxc_apparmor -
> >>lsm/apparmor.c:apparmor_process_label_set:171 - in your container
> >>configuration file
> 
> >Stricly speaking this is not an regression but an improvement.
> >1.0.x did silently ignore apparmor if you did not have "complete" apparmor 
> >support in the kernel (read: no mount restrictions which are only available 
> >in Ubuntu).
> >See [1] for details about the behaviour.
> >
> >With 1.1.x LXC actually errors out when it detecs you want apparmor but 
> >don't have the right kernel for it. You can set "lxc.aa_allow_incomplete = 
> >1" to allow it to use whatever apparmor support is available, skipping the 
> >cool mount stuff.
> 
> >That said I would love to close this as not-a-bug, if you do not disagree.
> 
> There was no news in apt-listchanges.
> I did find how to fix the problem googling the error and I had to understand
> what I was really doing adding that option but I wouldn't have incurred in
> the problem if:
> - the package itself would have made the change to the configurations (not
> sure if it is a good idea)
> or
> - there was a notice in apt-listchange.
> 
> I'm ok if you close the bug but I'm thinking about the other poor souls that
> may upgrade lxc and find their containers not running without any notice.

Fair enough, I will add a notice to the NEWS file to be picked up by 
apt-listchanges.

> Nothing terrible, I'm aware I'm using sid and probably once there will be a
> new kernel this won't be an issue.

When the feature will get upstream, yeah. I am waiting for this too.

Greets

Reply via email to