launchpad-dev team mailing list archive
-
launchpad-dev team
-
Mailing list archive
-
Message #07521
Re: ec2 seems horked (html5browser not importable)
-
To:
launchpad-dev@xxxxxxxxxxxxxxxxxxx
-
From:
Ian Booth <ian.booth@xxxxxxxxxxxxx>
-
Date:
Tue, 05 Jul 2011 21:30:16 +1000
-
In-reply-to:
<1309856551.9785.6.camel@solstice>
-
User-agent:
Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.2.17) Gecko/20110516 Lightning/1.0b2 Thunderbird/3.1.10
On 05/07/11 19:02, Curtis Hovey wrote:
> On Tue, 2011-07-05 at 09:18 +1000, Ian Booth wrote:
>>
>> Ah sorry. The seg fault happens when I run the yui tests inside a 32
>> bit
>> LXC container on a 64 bit OS. I misremembered the exact setup which
>> causes the issue.
>
> Does the LXC use the 64bit libs? html5browser is using a heuristic rule
> to decide if the calls must be in a subprocess:
> if arch == '64bit' and dist in ('natty', 'oneiric'):
> I can image this will fail in the container if the libwebkitgtk-*.so is
> actually the 64bit one.
>
> BTW. why a 32bit LXC? Wouldn't a 64bit lucid be more like the production
> env?
I'm running a 32 bit lucid container so the above if condition will fail
regardless of the bitness since dist is not natty or oneric.
I'm using 32 bit because it was suggested at the time of the original
discussions in the list that this would use far less memory compared
with a 64 bit setup.
I've gone back to running lp directly on my 64 bit natty system. I'm not
sure who else would be doing what I did in running a 32 bit container on
a 64 bit system. If there are others with the same problem, we should
try and solve it, but if it's just me.....
BTW. The rabbit test layer also fails inside LXC.
Follow ups
References