Ladies and Gentlemen, your food for thought.
Ran into a rather odd problem with RC1, trying to see how/if we can
recreate it currently. Consider this email just throwing out the flag for
everyone to hash over....
Setup:
FreeBSD 4.2-STABLE (from March 1st)
Apache 1.3.17
mod_perl 1.24
PHP 4.0.5RC1
Configure lines:
configure --with-imap \
--with-gettext \
--with-apxs=/usr/local/sbin/apxs \
--prefix=/usr/local \
--with-pgsql \
--enable-debug \
--enable-snmp \
--enable-track-vars \
--enable-trans-sid \
--disable-magic-quotes \
--enable-short-tags \
--without-mysql \
--without-tsrm-pthreads \
--enable-wddx
Error Type: Pages stopped being served. Massive spew of memory leaks in
httpd-error.log
Cause: Unknown at this time. We were doing some "stress" testing of our
setup, and running a system with concurrent 200-300 hits. Each hit was
running through a series of steps to simulate an actual user (tab
tab, return, tab, return... etc). It was running fine for over 24 hours,
but sometime during the night it seems things went south rather quickly.
What we do know is what appeared in the httpd-error.log and it looks like
so:
[Wed Mar 14 01:40:30 2001] [error] server reached MaxClients setting,consider raising
the MaxClients setting
./zend_execute.c(1487) : Freeing 0x0812D464 (12
bytes),script=/usr/local/www/data/error/error_http.php
zend_opcode.c(304) : Freeing 0x08175024 (4380
bytes),script=/usr/local/www/data/error/error_http.php
./zend_language_scanner.l(1138) : Freeing 0x081339A4 (5
bytes),script=/usr/local/www/data/error/error_http.php
Last leak repeated 25 times
zend_compile.c(812) : Freeing 0x0815C2E4 (18 bytes),
script=/usr/local/www/data/error/error_http.php
zend_variables.c(127) : Actual location (location was relayed)
./zend_language_scanner.l(1146) : Freeing 0x0815CDA4 (18
bytes),script=/usr/local/www/data/error/error_http.php
Last leak repeated 2 times
./zend_language_scanner.l(1337) : Freeing 0x0813A5E4 (7
bytes),script=/usr/local/www/data/error/error_http.php
.... etc etc ....
The same error repeated well until about 7AM this morning when someone
decided to restart the Apache server.
You can find a copy of the error log at
http://www.deadmime.org/~dank/httpd-error.log it only contains the
section from when things started to go wrong, until the restarting of
Apache (which fixed things it seems). The next earliest entry is from
March 2nd, so I don't think the two are/were related.
I've currently cvs updated to the latest 4.0.5RC1 (in case there was
anything out of date) and will begin testing this again, but don't expect
many results for at least 24 hours (earlier if we're (un)lucky).
Suggestions? Comments? Things I should try to remedy this?
I'd join onto the IRC and chat this up, but I'm seeming to have difficulty
joining any servers other than irc.mit.edu (and no one is in that #php).
Dan Kalowsky "Tonight I think I'll walk alone,
Worldgate Communications I'll find my soul as I go home."
Software Engineer - TICS Group - Temptation
--
PHP Development Mailing List <http://www.php.net/>
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
To contact the list administrators, e-mail: [EMAIL PROTECTED]
