The apache2 server frequently stops with the error below. (When I manually (re)start the apache2 server, it starts okay.)
The error messages in the apache2 log:
[mpm_prefork:notice] [pid 19521] AH00171: Graceful restart requested, doing restart
[:notice] [pid 19521] mod_python: Creating 8 session mutexes based on 150 max processes and 0 max threads.
[:notice] [pid 19521] mod_python: using mutex_directory /tmp
Fatal Python error: PyEval_SaveThread: NULL tstate
[core:notice] [pid 19521] AH00060: seg fault or similar nasty error detected in the parent process
[core:notice] [pid 19521] AH00060: seg fault or similar nasty error detected in the parent process
do you have any entries in the request log that might help you reproduce the error under controlled conditions? My gut feeling points at a problem in some lib that accessed via Python code (executed upon an HTTP request), causing that segmentation violation.
You may additionally want to enable core dumps for your httpd, so that you have a chance to look at the segv location post-mortem.