Oh sorry for the delay Matteo, you are always such a great help.
PHP Memory for our webservers is set to 512M per web node (we have 3).
The application is set to use 2048M. The application value was set before we went live with our load balanced farm so I'm wondering if that should be brought back down. I can look into ...
We are still load balancing with F5 and using dedicated memcached server for application cache (not sessions) and gluster software.
Performance has been really good since we implemented this set up 2 years ago.
Oh.. and we still use MySQL windows fail over cluster.
Can I send you a couple of crash reports? I'm not sure if I should be uploading that detailed information here.
Our linux admin thought the problem could be linked to oracle. We use the External Database plug-in for our automated course and enrolment set up so we have the oracle client installed to communicate with MySQL. This was configured years ago. I think I will disable this temporarily to see if there is any correlation.
I mean no regular updates on the same major but:
- PHP 7.x: 7.1 at least
- CentOS 7