1

I'm using a rather complex structure with multiple ssi-includes, set- and echo-commands.

The first document writes a lot of set-commands, includes another document which then again includes a third document. On the last included document the variable values are printed using the echo-command.

I noticed that with an increasing number of variables the probability for a segmentation fault to happen rises.

Did anyone experience something similar? How do I go about debugging such a problem?

I'm using IBM_HTTP_Server/2.0.47.1-PK65782 Apache/2.0.47

Benedikt
  • 123
  • 4

2 Answers2

1

Generally speaking Apache should never segmentation fault. I would use a test server and turn on all logging, and also peek into the process with strace to see where it is segfaulting at.

If you are sure it is mod_include, make sure it is compiled with the same version of apache you are running - if it was compiled for a different version that could explain the segfaults.

Dave Drager
  • 8,315
  • 28
  • 45
0

If it segfaults, you should be able to retrieve a core dump that will help to understand what happened at code level. Consult your operating system documentation to figure out how to collect the dump. Once you have it, you can examine the full internal state of the process at the time of the crash using your favourite debugger.

Strace doesn't help much with segfaults as it will only tell you what happens between userland and kernel while the segfault is usually indicative of a purely userland issue. If you are running Linux, valgrind and ltrace are more appropriate in this case. But first, get a backtrace out of the core.