I recently upgraded a production server from PHP 7.4 to PHP 8.0 (this is using the ondrej packages). As a reference, I'm using Ubuntu 20.04.3, apache2.4.46 and now PHP8.0.
After about 18 hours, the apache server stopped responding, giving an empty response data error. Apache gave the following error:
[core:notice] [pid 1285919] AH00051: child pid 1329445 exit signal Segmentation fault (11), possible coredump in /etc/apache2
Unfortunately no coredumps were in /etc/apache2/. I also saw conflicting information on enabling coredumps, namely that the CoreDumpDirectory could cause issues with file sizes and other things.
I was wondering a couple of things:
- How should I go about debugging apache2 on a production system? Should I enable coredumps? What is the best way to ensure it doesn't go crazy and fill up disk space?
- How can I ensure that it doesn't go down again? Is there some monitoring tool or tool I should use to restart apache?
- Is this just a PHP8.0 package issue? Has anyone run into this recently? Below are the packages installed via php -m:
[PHP Modules]
amqp
calendar
Core
ctype
curl
date
decimal
dom
exif
FFI
fileinfo
filter
ftp
gd
gettext
gmagick
hash
iconv
igbinary
intl
json
ldap
libxml
mbstring
memcache
memcached
mongodb
msgpack
mysqli
mysqlnd
OAuth
openssl
pcntl
pcre
PDO
pdo_mysql
pdo_sqlite
Phar
posix
raphf
readline
redis
Reflection
rrd
session
shmop
SimpleXML
sockets
sodium
SPL
sqlite3
standard
sysvmsg
sysvsem
sysvshm
tidy
tokenizer
uuid
xhprof
xml
xmlreader
xmlwriter
xsl
yac
yaml
Zend OPcache
zip
zlib
zmq
[Zend Modules]
Zend OPcache
I was able to just run service apache2 restart to get everything back up and running, but in the ~3 years or so from when I've done upgrades from PHP7/7.1/7.2/7.3/7.4, I haven't had any apache2 issues like this.
Any advice is appreciated.
