On Mar 3, 2008, at 1:10 PM, Christian Boos wrote: > > Some updates on the elusive #6614 bug > > Jeroen Ruigrok van der Werven wrote: >>>> I mean, might it make sense to release 0.11.0 now to get the >>>> release out the >>>> door and continue to look for these more elusive cases for a >>>> 0.11.1 release? >>>> >>> Well, it's a more a matter of fixing the Genshi issue for 0.5, >>> then make >>> Trac 0.11 depend on that version. As I said in the comment:38 >>> (linked >>> above), 0.4.x has the #G158 bug which makes Trac unreliable in >>> multithreaded setup, so I wouldn't recommend using it. >>> >> >> Ok, but then releasing Genshi 0.5 becomes a blocking priority before >> releasing Trac 0.11, since we need to get the eggs, especially for >> Windows >> with speedups, out there. >> > > Thanks to Alec who managed to write a Genshi-only test program > exhibiting the problem, we could narrow down the specific issue about > the memory constantly growing when an exception was raised from a > template to a Genshi bug. Well, I have the feeling it must rather be a > Python bug,
I'm not sure if this is related at all, since you're talking about 0.11 and I'm running 0.10-stable, but I have had a 2-year issue with memory and cpu usage getting out of control at dsource.org. With some help, gdb, and Python compiled w/ symbols, we tracked all the activity to _sre.c file (Python 2.4.4 source, but issue is in 2.5 as well). This is the regexp portion of Python and it seems to go into an endless loop. CPU spikes and the system starts swapping. I actually have a cronjob that restarts apache if utilization is too high... :( We didn't figure out what part of Trac triggered this within Python, but I thought I'd mention it for this thread. I had always assumed it was my use of a CentralDatabaseManager customization that I've made to the Trac code base. You're talking about high memory usage, and not so much CPU, so I don't know if it's related or not... > but at least it's triggered at (and can be worked around at) > the Genshi level. See #G190. > > But that issue is not the only problem we have w.r.t. to high memory > usage. I think it's still relatively easy to make the memory usage > grow > to a high number where it will stay forever, even with Python 2.5. > That > has to be improved further. > >>>> I am just trying to get a sense of the choice between finally >>>> releasing 0.11 >>>> and stability for the end-users. >>>> >>> Perhaps we can make another beta after the two patches on #6614 >>> (listed >>> in 1. and 2.a above) are committed, and see how things work in >>> practice? >>> >> >> Sure, I think our encountered issues warrant another beta regardless. >> >> > > The current state of affair is however quite better than it used to > be, > so we can probably prepare a beta2 even if #6614 is not completely > fixed > at this point. Thoughts? > > -- 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 -~----------~----~----~----~------~----~------~--~---
