← Back to team overview

launchpad-dev team mailing list archive

Re: possible regression on bug 672371

 

On Wednesday 10 November 2010 04:35:27 Robert Collins wrote:
> On Wed, Nov 10, 2010 at 1:42 PM, Robert Collins
> 
> <robertc@xxxxxxxxxxxxxxxxx> wrote:
> > So, this page has been a headache single pg 8.4 - core queries it runs
> > got massively slower, and it also does thousands of queries.
> > 
> > A fix for timeouts in getBuildRecordsForSourceIds in rev 15888 (which
> > works) has pushed this page out again, and we're seeing multiple
> > reports of timeouts on the page. We've given it a 20 second timeout,
> > but thats possibly not enough.
> > 
> > I'm going to see if I can get a solid fix up for review/landing before
> > EOD today. Would deeply appreciate someone in eu/us tz either
> > sheparding and deploying this fix (if I get one together), or
> > reverting and deploying if I don't.
> 
> I've confirmed that rev 15888 causes a regression in this page. The
> cause of the regression is previously decorator-based-preloading being
> subtly broken by the improvements to batch load at a lower layer. A
> candidate fix is up at lp:~lifeless/launchpad/soyuz -
> https://code.launchpad.net/~lifeless/launchpad/soyuz/+merge/40499
> 
> I'm leave the next two days, so I'm -really- hoping someone will step
> up and run with this - rollback or forward, either is fine with me.

It won't be me unfortunately since I'm fighting the buildd-manager breakage 
and on a deadline.  However Jelmer has stepped up and will get this landed, 
thanks Jelmer.



Follow ups

References