httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ivan Richwalski <i...@seppuku.net>
Subject Re: [STATUS] (apache-1.3) Fri May 1 23:45:24 EDT 1998
Date Tue, 05 May 1998 18:59:21 GMT
> On Sun, May 03, 1998 at 09:17:12PM -0700, Brian Behlendorf wrote:
> > By the way, I'm almost positive at this point that my core dumps were due
> > to using -O3 instead of just -O2 with gcc.  I'm now running with almost all
> > modules compiled in, and it's rock solid.
>
> Hmmh. Then I'm the only one with regular (and *very* strange) core dumps.
>     [Fri May  1 21:14:26 1998] [notice] httpd: caught SIGSEGV, attempting
>to dump core in /var/WWW/logs
>     [Fri May  1 21:14:41 1998] [notice] httpd: caught SIGSEGV, attempting
>to dump core in /var/WWW/logs
>     [Fri May  1 21:15:30 1998] [notice] httpd: caught SIGSEGV, attempting
>to dump core in /var/WWW/logs
>     [Fri May  1 21:17:09 1998] [notice] httpd: caught SIGSEGV, attempting
>to dump core in /var/WWW/logs
>     [Fri May  1 21:18:55 1998] [notice] httpd: caught SIGSEGV, attempting
>to dump core in /var/WWW/logs
>     [Fri May  1 21:19:14 1998] [notice] httpd: caught SIGSEGV, attempting
>to dump core in /var/WWW/logs
>     [Fri May  1 21:21:40 1998] [notice] httpd: caught SIGABRT, attempting
>to dump core in /var/WWW/logs
>     [Sat May  2 01:17:00 1998] [notice] httpd: caught SIGSEGV, attempting
>to dump core in /var/WWW/logs
>     [Sat May  2 23:28:04 1998] [notice] httpd: caught SIGSEGV, attempting
>to dump core in /var/WWW/logs
>     [Sun May  3 01:54:00 1998] [notice] httpd: caught SIGSEGV, attempting
>to dump core in /var/WWW/logs
>     [Mon May  4 09:51:17 1998] [notice] httpd: caught SIGSEGV, attempting
>to dump core in /var/WWW/logs
>     [Mon May  4 11:48:05 1998] [notice] httpd: caught SIGSEGV, attempting
>to dump core in /var/WWW/logs
> This is a server with a very *low* traffic (so it has a chance to timeout
> once in a while). I still suspect timeouts to trigger the dumps.

	I'm also seeing some strange semi-random core dumps occuring
on our web server too.  I have all of the modules compiled as shared
libs, and only load the ones we need at run-time.  I haven't recompiled
yet with -DPOOL_DEBUG and -DALLOC_DEBUG, but here's the how things are
setup now:

	SunOS:uname -a
	SunOS yima 5.6 Generic sun4u sparc SUNW,Ultra-1

	SunOS:gcc --version
	2.7.2.3

	SunOS:~http/sbin/httpd -V
	Server version: Apache/1.3b6
	Server built:   Apr 28 1998 17:04:14
	Server's Module Magic Number: 19980413
	Server compiled with....
	 -D HTTPD_ROOT="/usr/local/web"
	 -D HAVE_MMAP
	 -D USE_MMAP_SCOREBOARD
	 -D USE_MMAP_FILES
	 -D USE_FCNTL_SERIALIZED_ACCEPT

	SunOS:grep "^LoadModule" conf/httpd.conf
	LoadModule      access_module           lib/mod_access.so
	LoadModule      action_module           lib/mod_actions.so
	LoadModule      alias_module            lib/mod_alias.so
	LoadModule      auth_module             lib/mod_auth.so
	LoadModule      db_auth_module          lib/mod_auth_db.so
	LoadModule      autoindex_module        lib/mod_autoindex.so
	LoadModule      cgi_module              lib/mod_cgi.so
	LoadModule      dir_module              lib/mod_dir.so
	LoadModule      expires_module          lib/mod_expires.so
	LoadModule      imap_module             lib/mod_imap.so
	LoadModule      includes_module         lib/mod_include.so
	LoadModule      info_module             lib/mod_info.so
	LoadModule      config_log_module       lib/mod_log_config.so
	LoadModule      mime_module             lib/mod_mime.so
	LoadModule      negotiation_module      lib/mod_negotiation.so
	LoadModule      setenvif_module         lib/mod_setenvif.so
	LoadModule      status_module           lib/mod_status.so
	LoadModule      userdir_module          lib/mod_userdir.so

	SunOS:/usr/local/web/sbin/apachectl status

	                    Apache Server Status for gb.marcomm.com

	   Server Version: Apache/1.3b6
	   Server Built: Apr 28 1998 17:04:14
	  _________________________________________________________________

	   Current Time: Tue May 5 12:59:41 1998
	   Restart Time: Fri May 1 09:45:03 1998
	   Server uptime: 4 days 3 hours 14 minutes 38 seconds
	   Total accesses: 411795 - Total Traffic: 2.0 GB
	   CPU Usage: u8.48 s2.43 cu9.63 cs3.65 - .00677% CPU load
	   1.15 requests/sec - 6.0 kB/second - 5.2 kB/request
	   13 requests currently being processed, 8 idle servers




	I've been trying to catch the problem using truss to trace all of
the system calls that the server uses with this command:
	truss -o httpd.log -r all -w all -fp `cat logs/httpd.pid`

	The logs are a bit lengthy, so instead of posting them I'll put
them up on the web:
	<http://www.seppuku.net/~ivan/sigbus-log.txt>
	<http://www.seppuku.net/~ivan/sigsegv-log.txt>

Ivan.

Mime
View raw message