Apache Segmentation Fault
- Date: Fri, 23 Nov 2018 16:38:39 +0100
- From: "H." <work@xxxxxxxxx>
- Subject: Apache Segmentation Fault
I have a web server running Testing with apache 2.4.37 and php 7.3. The
purpose of the server is to run owncloud. However, after the nightly
restart the server stops working. The logs show the following:
[Fri Nov 23 00:00:01.526101 2018] [mpm_prefork:notice] [pid 2858]
AH00171: Graceful restart requested, doing restart
[Fri Nov 23 00:00:01.622137 2018] [mpm_prefork:notice] [pid 2858]
AH00163: Apache/2.4.37 (Debian) OpenSSL/1.1.1 configured -- resuming
[Fri Nov 23 00:00:01.622156 2018] [core:notice] [pid 2858] AH00094:
Command line: '/usr/sbin/apache2'
[Fri Nov 23 00:00:01.626193 2018] [core:notice] [pid 2858] AH00052:
child pid 5383 exit signal Segmentation fault (11)
[Fri Nov 23 00:00:01.626217 2018] [core:notice] [pid 2858] AH00052:
child pid 5384 exit signal Segmentation fault (11)
[Fri Nov 23 00:00:01.626222 2018] [core:notice] [pid 2858] AH00052:
child pid 5385 exit signal Segmentation fault (11)
[Fri Nov 23 00:00:01.626227 2018] [core:notice] [pid 2858] AH00052:
child pid 5386 exit signal Segmentation fault (11)
These segmentation faults keep going so that the error.log file grows
to hundreds of MB within a few hours.
When I do "service apache2 restart" the server works properly again
until the next nightly restart.
Any idea what could cause this problem?
libapache2-mod-php7.3/testing,now 7.3.0~rc4-1 amd64
php-apcu/testing,now 5.1.12+4.0.11-2 amd64
php-redis/testing,now 4.2.0~rc2-1 amd64
php7.3-opcache/testing,now 7.3.0~rc4-1 amd64
certbot/testing,now 0.28.0-1 all
First I had the opcache enabled in /etc/php/7.3/apache2/php.ini, but
then I disabled it since many people reported issues with opcache
causing segfaults. However, the problem still persists. I also tried to
switch the memcache setting of owncloud from APCu to Redis, but also no
effect. Owncloud's cron.php is executed via www-data's crontab.
I also ran "certbot renew", but it does nothing because the server's
certificate is still valid and thus the server keeps running.