Jonas Borgström wrote: > Christian Boos wrote: > >> Hello, >> >> There's a patch I'd like to commit on stable before 0.11.1, but first >> I'd like to check if nobody foresees a problem with this, and also if I >> could get some testing feedback, particularly from people still >> experiencing db timeout errors. >> >> This patch should improve upon concurrency and reduce the risks for a >> timeout while waiting for a db connection, when Trac is used with a >> cached svn repository (i.e. probably still the typical situation). >> ... >> >> It'd be nice if jonas could try it on some heavily used Trac ;-) >> > > The patch has been running on t.e.o for a couple of hours now and > everything seems to be working ok. There might also be slightly fewer > active database connections now but it's hard to tell since Saturdays > are the least busy day of the week anyway. >
Thanks for testing it. I applied the patch in [7375], correcting a typo in sync_changeset() (there's no post-revprop-change hook doing a "trac-admin resync <rev>" on t.e.o, so that error couldn't have been harmful). Let's see how it goes. The only problem I can imaging is that by enabling more reuse of the connections, this could uncover problems with sharing connections between threads. We still don't allow that for Pysqlite on anything but windows, so there's not much risk there. Maybe concurrency issues for older psycopg2 bindings? Anyway, if some configuration proves to be problematic, better disable it by proper use of the poolable flag, as that will be more robust than simply relying on less reuse. -- Christian --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Trac Development" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/trac-dev?hl=en -~----------~----~----~----~------~----~------~--~---
