Environment:
- FreeBSD i386 4.9 (MAX_USERS=512)
- Load averages: 0.73, 0.83, 0.93
- Mem: 727M Active, 712M Inact, 251M Wired, 85M
Cache, 199M Buf, 236M Free (2GB physical)
- Swap: 1059M Total, 2896K Used, 1056M Free
- Apache 1.3.27 (5 con-current, working on different
IP's)
- PHP 4.2.3 (one php.ini for all Apache's, including
MMCache's own config)
- MMCache 2.4.6
__________________________________
``ipcs -T`` output:
shmmax: 67112960 (max shared memory segment
size)
shmmin: 2 (min shared memory segment size)
shmmni: 257 (max number of shared memory
identifiers)
shmseg: 65 (max shared memory segments per
process)
shmall: 16385 (max amount of shared memory in pages)
__________________________________
``netstat -m`` output:
479/1328/34816 mbufs in use (current/peak/max):
479 mbufs allocated to data
396/1078/8704 mbuf clusters in use (current/peak/max)
2488 Kbytes allocated to network (9% of mb_map in use)
0 requests for memory denied
0 requests for memory delayed
0 calls to protocol drain routines
__________________________________
Turck MMCache's configuration:
[mmcache]
mmcache.admin.name="mmcache"
mmcache.admin.password="xxxxxxxxxxxxxxx"
zend_extension="/www/modules/mmcache.so"
mmcache.shm_size="32"
mmcache.cache_dir="/var/tmp/mmcache"
mmcache.enable="1"
mmcache.optimizer="1"
mmcache.check_mtime="1"
mmcache.debug="0"
mmcache.filter=""
mmcache.shm_max="0"
mmcache.shm_ttl="7200"
mmcache.shm_prune_period="0"
mmcache.shm_only="1"
mmcache.compress="0"
mmcache.keys="shm_only"
mmcache.sessions="disk_only"
mmcache.content="disk_only"
__________________________________
With the above configuration, which to me looks like
wonderful, Apache still crashes here and there with sigal
11 errors. Needless to say that turning off MMCache
completely eliminates those errors.
A snippet from ``/var/log/messages``:
Dec 17 14:26:40 dns /kernel: pid 82868 (httpd1), uid
10000: exited on signal 11
Dec 17 14:26:40 dns /kernel: pid 83366 (httpd2), uid
10000: exited on signal 11
Dec 17 14:26:41 dns /kernel: pid 82913 (httpd1), uid
10000: exited on signal 11
Dec 17 14:26:41 dns /kernel: pid 83386 (httpd5), uid
10000: exited on signal 11
Dec 17 14:26:43 dns /kernel: pid 82912 (httpd1), uid
10000: exited on signal 11
Dec 17 14:26:43 dns /kernel: pid 83385 (httpd5), uid
10000: exited on signal 11
Dec 17 14:26:44 dns /kernel: pid 83398 (httpd2), uid
10000: exited on signal 11
Dec 17 14:26:44 dns /kernel: pid 83384 (httpd5), uid
10000: exited on signal 11
Dec 17 14:26:45 dns /kernel: pid 82901 (httpd1), uid
10000: exited on signal 11
Dec 17 14:26:46 dns /kernel: pid 82858 (httpd1), uid
10000: exited on signal 11
Dec 17 14:26:46 dns /kernel: pid 83397 (httpd2), uid
10000: exited on signal 11
Dec 17 14:26:46 dns /kernel: pid 83383 (httpd5), uid
10000: exited on signal 11
Dec 17 14:26:47 dns /kernel: pid 82919 (httpd1), uid
10000: exited on signal 11
Dec 17 14:26:48 dns /kernel: pid 82857 (httpd1), uid
10000: exited on signal 11
Dec 17 14:26:48 dns /kernel: pid 83395 (httpd2), uid
10000: exited on signal 11
Dec 17 14:26:48 dns /kernel: pid 83367 (httpd2), uid
10000: exited on signal 11
Dec 17 14:26:48 dns /kernel: pid 83379 (httpd5), uid
10000: exited on signal 11
Dec 17 14:26:50 dns /kernel: pid 83378 (httpd5), uid
10000: exited on signal 11
Dec 17 14:26:51 dns /kernel: pid 83352 (httpd2), uid
10000: exited on signal 11
Dec 17 14:26:52 dns /kernel: pid 83380 (httpd5), uid
10000: exited on signal 11
Dec 17 14:26:53 dns /kernel: pid 83409 (httpd1), uid
10000: exited on signal 11
Dec 17 14:26:53 dns /kernel: pid 83311 (httpd2), uid
10000: exited on signal 11
Dec 17 14:26:53 dns /kernel: pid 83373 (httpd5), uid
10000: exited on signal 11
Dec 17 14:26:54 dns /kernel: pid 83416 (httpd2), uid
10000: exited on signal 11
Dec 17 14:26:54 dns /kernel: pid 83396 (httpd4), uid
10000: exited on signal 11
Dec 17 14:26:54 dns /kernel: pid 83226 (httpd5), uid
10000: exited on signal 11
(Notice dates and times... every second or so, another
process crashes.)
Logged In: YES
user_id=708289
I had same problems, I backed out to version 2.4.1 and it
works OK, the only problem now is when I do 'apachect
restart' it gives me segmentation faults errors (signal 11) but
that's not much important - it's not too often.
After disabling mmcache, I haven't recorded any signal 11
errors for more than 1 month...
My config:
- FreeBSD i386 4.7
- Mem: 224M Active, 580M Inact, 133M Wired, 32M Cache,
112M Buf, 34M Free (1GB physical)
- Swap: 2048M Total, 1572K Used, 2046M Free
- Apache/1.3.28 (Unix) PHP/4.3.4 mod_ssl/2.8.15
OpenSSL/0.9.7c
- MMCache 2.4.6
-----------------
netstat -m
261/800/26624 mbufs in use (current/peak/max):
261 mbufs allocated to data
259/738/6656 mbuf clusters in use (current/peak/max)
1676 Kbytes allocated to network (8% of mb_map in use)
0 requests for memory denied
0 requests for memory delayed
0 calls to protocol drain routines
-------------
[mmcache]
zend_extension="/usr/local/lib/php/20020429/mmcache.so"
mmcache.shm_size="16"
mmcache.cache_dir="/tmp"
mmcache.enable="1"
mmcache.optimizer="1"
mmcache.check_mtime="1"
mmcache.debug="0"
mmcache.filter="!/usr/local/www/horde/*"
mmcache.shm_max="0"
mmcache.shm_ttl="0"
mmcache.shm_prune_period="0"
mmcache.shm_only="0"
mmcache.compress="0"
mmcache.keys="shm_and_disk"
mmcache.sessions="shm_and_disk"
mmcache.content="shm_and_disk"
Logged In: YES
user_id=701927
This also happens on our server:
Dual CPU 1Ghz
1 GB Ram
Freebsd 4.5
Apache 1.3.29 / mod_ssl
Php 4.3.4
Logged In: NO
My sparc solaris machines have same problem. but linux boxes
are O.K.
My sparc machines use Solaris 8 and Solaris 2.6.
Logged In: YES
user_id=961290
I have same problem with sparc solaris boxes. my sparc
machines use Solaris 8 and Solaris 2.6.
My linux boxes have no problem such like this.
Logged In: NO
This happens on my linux system too..
HINT: for me it only happens after at least ONE apache
RELOAD. It never happens if I STOP/START apache!
In the case when it happens it seems that it happens after
serving the request before the child process exits
Logged In: YES
user_id=876499
I see the same here with PHP 4.3.4/mmcache 2.4.1 (RedHat 7.3)
It wasn't seen with PHP 4.1.2/mmcache 2.4.1
Logged In: NO
Recompile php-mmcache if you have such warning after starting
apache. I had similar situation after upgrading apache without
recompiling it...
Logged In: YES
user_id=974724
This also happens on our server:
Dual CPU 800Mhz
1 GB Ram
Unixware 7.1.1
Apache 1.3.28
Php 4.3.4
Any patch?
Logged In: YES
user_id=976053
I use Linux
(several 1 CPU servers with different kernel, Apache 1.3)
PHP 4.3.4 MMCache 2.4.6 and have the similar problem.
I have Apache crashes (segfault) about once a day.
I guess it happens under the high load, when two or more
HTTP requests hit the same page at the same time.
I use only code caching:
mmcache.shm_only="1"
mmcache.optimizer="1"
No Apache RESTART, everything recompiled.
Any workaround?
--------------------
Best Regards, Vladimir
Logged In: YES
user_id=974724
We put PHP 4.3.4 and MMCache 2.4.1 and all problems seems
to desappear.
Logged In: NO
Same problem with PHP5-beta4-snapshot-200422* on a
redhat 9.
BUT !
In my root directory i've got .htaccess with an
php_auto_prepend_file utf8decode.php
When i enable it -> segfault. Disable ->ok. I must restart
apache between each test.
Logged In: NO
i have the same problem on solaris 8 (sun4u), apache 1.3.29,
php 4.3.4, mmcache 2.4.6.
(same problem on 8 similar machines)
after an apache restart, i have no problems, 2-3 hours later
(depending on user-traffic), it starts to segfault.
i simple apache stop-start resolves it for the next hours...
workaround? in my case a cronjob grep for segfaults in a
tail on error_log. if 10 out of 20 are segfaults and no
apache sigterm message is found, stop, sleep, start.
ugly, but working...
Logged In: NO
I have this problems on OpenBSD
apache: 1.3.28
php: 4.3.4
mmcache: 2.4.6
Logged In: YES
user_id=1019025
same here, on linux with php 4.2.2 mmcache 2.4.6 on a dual
opteron system. usually works fine for a few hours before the
segmentation faults appear. apache restart helps again for a
few hours. Thanks for any help on this issue.
Logged In: NO
i had the same problems under solaris 8 & solaris 9.
the only intersting entries in logfiles whenever it started
to segfault was a higher amount of concurrent requests (more
than "minspareservers" apache settings).
so i increased the minspareservers & maxspareservers
settings - and the problem has gone.
it is not a solution - and there are not that much logical
reasons why this workaround really should work - but it was
the only change i made to get it stable.
Logged In: NO
Same problems occuring on 4 FreeBSD servers 4.7-4.10.
Very similar vitals compared to the original bug report.
Logged In: YES
user_id=1069839
Same problems occuring on Solaris8 for SPARC.
Apache-1.3.29 and PHP-4.3.6 with MMCache-2.4.6.
Does anybody have a good idea for fix ???
My php.ini (for mmcache) configuration are following.
[mmcache]
zend_extension="/ul/lib/php/extensions/no-debug-non-zts-
20020429/mmcache.so"
mmcache.shm_size="16"
mmcache.cache_dir="/tmp/mmcache"
mmcache.enable="1"
mmcache.optimizer="1"
mmcache.check_mtime="1"
mmcache.debug="0"
mmcache.filter=""
mmcache.shm_max="1M"
mmcache.shm_ttl="3600"
mmcache.shm_prune_period="900"
mmcache.shm_only="0"
mmcache.compress="1"
And I got following error many many many times.
[Wed Jun 23 10:36:15 2004] [notice] child pid 14391 exit
signal Segmentation Fault (11)