Setting PHP's opcache parameter to 1 (on, the default) is giving me 'zend_mm_heap corrupted' errors in the Apache log file at a rate of a few a day, irregularly.
Previous StackOverflow answers have suggested this might be because of (a) using other caching modules such as APC - not the case here, as only the standard PHP distribution is being used without any non-native caching or (b) running out of memory - but I have at least 1.6G of swap space available according to free -m
or (c) a bug in the PHP compiler - unlikely as this error is not widely reported and I am not doing anything unusual.
The server runs several websites, built using Drupal, Joomla, and bespoke PHP. I am running a standard PHP5.6.36 with mod_php and Apache 2.4.33 using the event MPM on Amazon Linux 2.
The only thing that is not completely standard stuff is that I am using the Amazon AWS SDK for PHP v. 3 to send mail, but I have no reason to suppose that this is causing the problem.
How can I track down what is causing the heap corruption?