[users@httpd] Apache v1.3.27 ssl_mutex.XXYYZ segmentation fault ?

[Julian Gomez <ju...@fomema.com.my>]

Hi,

I have a webserver, with the following configuration :

	-> Apache v1.3.27
	-> Mod_SSL v2.8.14
	-> PHP v4.3.2

PHP is then built with MySQL v4.0 & Oracle 8i client libraries. PHP and
Mod_SSL was built as DSOs. It seems that I am getting an rather large
amount of 'segmentation fault (11)' entries in the 'error_log'. The
specific entry is :

[Thu Jul  3 15:42:55 2003] [notice] child pid 21886 exit signal
Segmentation fault (11)

[Thu Jul  3 15:43:07 2003] [notice] child pid 21904 exit signal
Segmentation fault (11)

[Thu Jul  3 15:44:38 2003] [notice] child pid 22353 exit signal
Segmentation fault (11)

Those are happening close to every 1-2 minutes apart. That said, it doesn't
seem to have screwed our SSL setup badly or anything like it. Or at least I
haven't heard anyone scream yet.

I have run 'strace -tttt -vvv -f -F -p (apache-ppid)' and the last couple
of lines of output from a segment-faulted child process is like :

[pid 16750] 1057283001.817245 close(7)  = 0
[pid 16750] 1057283001.817318 close(6)  = 0
[pid 16750] 1057283001.817351 brk(0)    = 0x814e000
[pid 16750] 1057283001.817386 brk(0)    = 0x814e000
[pid 16750] 1057283001.817416 brk(0x8114000) = 0x8114000
[pid 16750] 1057283001.817479 brk(0)    = 0x8114000
[pid 16750] 1057283001.817512 flock(23, LOCK_UN) = 0
[pid 16750] 1057283001.817555 --- SIGSEGV (Segmentation fault) @ 0 (0) ---
[pid 16750] 1057283001.817703 chdir("/usr/local/apache") = 0
[pid 16750] 1057283001.817753 rt_sigaction(SIGSEGV, {SIG_DFL}, {SIG_DFL},
8) = 0[pid 16750] 1057283001.817795 getpid()  = 16750
[pid 16750] 1057283001.817830 kill(16750, SIGSEGV) = 0
[pid 16750] 1057283001.817863 sigreturn() = ? (mask now [RTMIN])
[pid 16750] 1057283001.817919 --- SIGSEGV (Segmentation fault) @ 0 (0) ---
[pid 16679] 1057283001.845411 <... select resumed> ) = 1 (in [16])

Checking fd 23 under /proc/ for the corresponding PID points to the
'ssl_mutex.(ppid)' file. That file is _zero_ bytes in length. I've traced a
couple of other child httpd processes, and they all die the same way.

I've dug through the mailing list archives at MARC, and it doesn't seem to
have turned up anything specific. I am presuming that this should not be
happening since I have my webmail system served up with SSL, with a lower
load admittedly and haven't seen anything like this at all.

Thinking it was possibly a memory leak between Apache & PHP, I've restarted
Apache, but the errors reoccur after about 4-5 minutes. I cannot run
memtest86 on the system because its in production, but I have used it as my
desktop for about 2 weeks prior to this without issues.

Would compiling the above combination of software as DSOs in the wrong
sequence possibly cause anything like this ?

Anyone care to hit me over the head with a cluestick ? :)

Thanks!

---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
   "   from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org