launchpad-dev team mailing list archive

[Guilherme Salgado <salgado@xxxxxxxxxxxxx>]

On Mon, 2010-08-16 at 22:26 +1000, Steve Kowalik wrote:
> Hi,
> 
> 	Currently, we are seeing EC2 instances having the test runner die in a
> heap due to the OOM killer stepping in and killing the test runner since
> it's consuming all of the RAM.
> 
> 	I had a branch fall victim to this over the weekend, and nailed it down
> to lib/lp/soyuz/doc/buildd-slavescanner.txt, line 51, in the call to
> .updateStatus().
> 
> 	Robert has seen this impact on his own landing, so I have handed off to
> jml who will investigate further.

Sometime mid last week I made my Lucid AMI public, which means from then
on anybody using ./utilities/ec2 would use that AMI instead of the old
Hardy one.

I'm thinking of making the Lucid AMI private again just to see if the
same problem occurs on the Hardy one.  Once it's private I can grant
access to it on a per-developer basis so that you can debug this OOM
issue.  What do you all think?

-- 
Guilherme Salgado <https://launchpad.net/~salgado>

Attachment: signature.asc
Description: This is a digitally signed message part