On Wed, Aug 01, 2012 at 08:11:46PM +0200, Marco Sehrer [ninjaconcept.com] wrote: > > great, I gave it a try, our test-suite runs green. and as far as I can see, > it works correctly with multiple workers too. > > I took a manual benchmark with 30 workitems running thru a workflow: > > 2 workers - 1:45 ruote-redis > 2 workers - 1:37 ruote-redis-experimental > > But this is not very representative as we have some time-consuming jobs going > on, I think the pure ruote-speedup must be far more. > Would be cool, to have this optimization in master soon.
Hello Marco, I merged it into master. The number of pop attempts is configurable via a 'pop_count' option (defaults to 28). Maybe the next way to improve would be to have pool of threads for the participants so that one participant doesn't monopolize the whole worker. The number of threads could be limited (so we don't run out of database connections like you described previously). Just tell me if you'd be interested by such a refinement. I cannot do it immediately, I have to work on ruote-sequel for Chad. Cheers, -- John Mettraux - http://lambda.io/jmettraux -- you received this message because you are subscribed to the "ruote users" group. to post : send email to [email protected] to unsubscribe : send email to [email protected] more options : http://groups.google.com/group/openwferu-users?hl=en
