Description

At random (at least it seems), when I try to load a page from my wiki, an exception is caught by mod_python and a traceback is shown.

Steps to reproduce

Either it happens at random or I couldn't figure out a reliable way to reproduce it. It doesn't happen that much, but I just:

1. Go to any page on my wiki;
2. Try to edit it;
3. Hit 'Cancel'.
4. Repeat from second step until the error pops up.

Example

I've managed to open a non-sensitive area of my wiki: https://boreste.dyndns.org/OndeComer . Just to let you know, the SSL security certificate is invalid, don't worry. The server isn't up 24/7 - it is mostly on brazilian work hours - 12h GMT to 23h GMT.

• I can't reproduce it following the steps above. How many tries do you need (I have tried about 30-40 cycles)? Text editor or GUI editor? -- ThomasWaldmann 2006-03-21 17:34:22

Details

The traceback is not caught by MoinMoin. I get this error message (I guess generated by mod_python): traceback.txt

• I have applied the patch (404) and the problem persists. I have diff'ed the

  tracebacks - nothing different. -- CarlosVieira 2006-02-06 11:59:48

According to SystemInfo:

Workaround

Just keep reloading the page. Alternatively, if the page was a submit (i.e. contains POSTDATA), go back and forth. The problem may repeat itself a little bit but vanishes after a few (1 to 3) retries.

Discussion

Please also add the exact mod_python version you use.

• # rpm -qa |grep mod
  mod_auth_mysql-20030510-5
  mod_ssl-2.0.53-3.3
  mod_python-3.1.3-5.2

Does it also happen in a std cgi setup (without mod_python)?

• I'll be testing it.

  • The workload has grew too much, I won't be able to test it within the next months (!!) -- CarlosVieira 2006-02-04 23:31:38

If you have time, could you try using fastcgi as a mod_python alternative?

• I wouldn't like to go through the hassle of installing and configuring fastcgi it I can avoid it. I have no CGIs in use.

This looks like our race condition again. Of course the author of the failing code won't admit it. Cf. MoinMoinBugs/DeepCopyError -- AlexanderSchremmer 2006-01-26 21:17:31

It reads BugFixed, but where's the patch? By the way, could anyone reproduce it (or is it just me)? -- CarlosVieira 2006-01-27 10:34:12

• I have applied the patch, but got the same traceback (exactly the same, according to diff). It actually got a little worse - it wasn't until I reloaded apache config that the problem got away. I ask again: has anyone else seen it?

  Edited -- CarlosVieira 2006-02-06 11:59:48

I have this same problem, only since switching to mod_python 2 weeks ago, and it looks like I be switching back to cgi based on your information. -- JonElsonGreen 2006-03-29 22:06:00

I have a similar problem after upgrading from 1.3.5 to 1.5.3 - the traceback shows same PickleError problem as reported here. Hard to reproduce - it happens either at random or sometimes on every request, thus blocking ability to use the wiki. Restart may work as a temporary solution. See the traceback: traceback_captsolo.txt

• -- CaptSolo 2006-04-26 22:59:00

Fix!?

I had exactly this problem and I solved changing PythonHandler moinmodpy to PythonHandler MoinMoin.request::RequestModPy.run in apache configuration. I don't know why it is working, but I'm sure that this corrected the problem for me.

rodrigo dot rd atat gmail dot com 2006-05-29 14:40:48

• Please give your exact mod_python version. -- ThomasWaldmann 2006-05-29 20:38:58

  I'm experiencing this problem even when using PythonHandler MoinMoin.request::RequestModPy.run. System info: Python 2.3.3, mod_python 3.1.3, MoinMoin 1.5.2. It seems to occur more frequently when two wiki instances are in active use, although it's hard to tell for sure, since the problem is so sporadic. -- JasonPratt 2006-06-07 00:45:09

Plan

• Priority:
• Assigned to:

• Status: Fixed in the 1.6 branch, see the first hit here: http://hg.alexanderweb.de/moin-1.6-main?cmd=changelog&rev=wikidict+alexander

CategoryMoinMoinBugFixed