[Buildbot-commits] [SPAM] Re: [Buildbot] #829: Buildbot 0.8.0rc2 waterfall is unusably slow

[Buildbot]

#829: Buildbot 0.8.0rc2 waterfall is unusably slow
---------------------+------------------------------------------------------
Reporter:  bdbaddog  |       Owner:       
    Type:  defect    |      Status:  new  
Priority:  critical  |   Milestone:  0.8.0
 Version:  0.8.0     |    Keywords:       
---------------------+------------------------------------------------------

Comment(by dustin):

 I've been experimenting with this in the context of the metabuildbot.  It
 has 937 changes at the moment, and the waterfall renders in 1.147s, and is
 similarly quick even with a build running.

 What do the memory footprints of your (willw, bdbaddog) masters look like?
 Is it possible that this is so slow because they're swapping?

 My earlier idea that this might be caused by the waterfall yielding time
 to other activities on the master (like processing data from a slave
 performing a build) is roundly contradicted: the waterfall's
 {{{buildGrid}}} method is a monolithic, blocking method (it does use
 generators to get some multiprocessing going, but it doesn't allow the
 reactor to process other events).

-- 
Ticket URL: <http://buildbot.net/trac/ticket/829#comment:16>
Buildbot <http://buildbot.net/>
Buildbot: build/test automation