Subject: Re: [boost] [trac] Performance for last week?
From: Dean Michael Berris (mikhailberis_at_[hidden])
Date: 2010-12-06 21:29:36
On Tue, Dec 7, 2010 at 9:46 AM, Eric Niebler <eric_at_[hidden]> wrote:
> On 12/6/2010 5:59 PM, Rene Rivera wrote:
>> First after disabling the trac source browser for most of the week
>> here's the new breakdown of the top ten sections the web server spent
>> time on (including www.boost.org):
> Thanks for working on this, Rene.
>> So big questions with respect to last week:
>> * Is trac performance to an "acceptable" point?
> Slightly less horrible perhaps (hard to say), but I did get another
> "database is locked" error today while trying to view a changeset. I
> think it was this:
> Then when I click on one of the links there (e.g.
> I get an error. I guess that's because you removed that plug-in, is that
I just tried to make some changes myself to a ticket and I see that
the server's utilization is up again to 11.38 -- that's not good, but
when the anonymous ticket filing was turned off I didn't see the load
average number to be going above 1.x. Maybe we should just disable
anonymous issue posting?
Also I did get an error submitting a comment to an issue, but not the
database is locked error that you're getting Eric.
>> * Do you miss the trac source browser, and desperately want it back?
> I guess I do miss it. It's especially annoying that trac is serving up
> links that are now broken, and I'm still getting database-is-locked
> errors. Is there any way to profile page requests, see which are taking
> the longest, and where they're spending their time?
I personally don't miss it.
Eric, as far as profiling goes, I'm not positive that Trac has that
capacity. Although Python has a profiling module, running Trac under a
Python profiler would really kill performance. If you think it's
unacceptable now, then turning on profiling is going to be quite
Maybe we should get Google Analytics instead into the pages to at
least have a better insight into the usage scenarios when it comes to
browsers going through the site. This would give us end-user
statistics, which would also be useful to confirm/debunk assumptions
we're making about Trac usage.
-- Dean Michael Berris deanberris.com
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk