hi I'm running Zope 2.6.0 (instance_home) with Python 2.1.3 on FreeBSD 4.7. I applied this patch to Python (http://zwiki.org/PatchPython213) to increase the thread stack size as the instance in question runs a lot of wikis and Plone sites. Everything was fine for about 7-10 days but now i'm getting seemingly-random, repetitive and isolated python core dumps.
From looking at Apache logs, events.log and the undo "log" in the root of the instance, the only connection i can remotely make is that many requests were made at the moment it core dumped (and i can only say that for one of the times it core dumped.)
I'm running Plone beta 3 and zwiki 0.14.0. the ZODB is about 200MB. This is on a dual proc XEON machine with 2GB RAM. I know about this thread: http://mail.zope.org/pipermail/zope-dev/2002-September/thread.html#17416 But I'm not coming away with anything actionable other than perhaps apply a different patch to Python? Here's the output from gdb on the latest core dump: This GDB was configured as "i386-unknown-freebsd". Core was generated by `python2.1'. Program terminated with signal 10, Bus error. #0 0xa272b8c in ?? () Most other dumps were on signal 11 Can anyone suggest how i can go about interpreting this? Here is the output from /var/log/messages on the latest batch of core dumps: Feb 3 08:26:08 alpha /kernel: pid 84972 (python2.1), uid 1006: exited on signal 11 (core dumped) Feb 3 08:26:17 alpha /kernel: pid 92001 (python2.1), uid 1006: exited on signal 11 (core dumped) Feb 3 08:26:31 alpha /kernel: pid 92010 (python2.1), uid 1006: exited on signal 10 (core dumped) Feb 3 08:30:03 alpha /kernel: pid 92026 (python2.1), uid 1006: exited on signal 10 (core dumped) I'd appreciate any advice on how to go about analyzing this and finding a solution. thanks. <--> george donnelly - http://zettai.net/ - "We Love Newbies" :) Zope Hosting - Dynamic Website Design - Search Engine Promotion Yahoo, AIM: zettainet - ICQ: 51907738 - e:george@zettai.net
george donnelly wrote:
Here is the output from /var/log/messages on the latest batch of core dumps:
Feb 3 08:26:08 alpha /kernel: pid 84972 (python2.1), uid 1006: exited on signal 11 (core dumped) Feb 3 08:26:17 alpha /kernel: pid 92001 (python2.1), uid 1006: exited on signal 11 (core dumped) Feb 3 08:26:31 alpha /kernel: pid 92010 (python2.1), uid 1006: exited on signal 10 (core dumped) Feb 3 08:30:03 alpha /kernel: pid 92026 (python2.1), uid 1006: exited on signal 10 (core dumped)
I'd appreciate any advice on how to go about analyzing this and finding a solution. thanks.
If you have the cores from them try running gdb on the core to find out where it was when it died. It's still possible you ran out of stack depending on how much that patch added. I normally only make the stack about 128k per thread on BSD, and its possible that is a bit small. Bus errors imply that data to a system call was incorrect (not properly aligned, not mapped, etc).
participants (2)
-
george donnelly -
Matthew T. Kromer