Just recently I've started to see seemingly random crashes on a web server that has been stable for quite some time. I think PHP is crashing, but there are no crash logs so I'm only assuming that's what the lighttpd errors mean, that PHP is not running. I don't know what's changed, it could be the requests coming in or could be the PHP code.
We run Rocky Linux, lighttpd server, PHP v7.4.
Here's an excerpt of the lighttpd error logs. I don't see any other errors in messages or php error logs. Requests are coming in fine, then all of a sudden the error log is spammed with loads of these and all requests are receiving a HTTP 500 response, then it all starts working again a couple of mins later (assume the PHP process is auto started again).
(network_write.c.98) write failed: 137: Transport endpoint is not connected
(gw_backend.c.2065) connection was dropped after accept() (perhaps the gw process died), write-offset: 0 socket: unix:/var/run/php-fpm/www.sock
(http-header-glue.c.1342) read() 136 137: Invalid argument
(gw_backend.c.2419) response not received, request not sent on socket: unix:/var/run/php-fpm/www.sock for xxxx, reconnecting
How can I track down what's causing this issue?