Resolved nginx random bad gateway 502 errors segfault

[finbarr69]

Server operating system version

CentOS Linux release 7.9.2009 (Core)

Plesk version and microupdate number

Plesk Obsidian Version 18.0.44 Update #3

Working on a Joomla 4 website today and I get random Nginx Bad Gateway 502 errors, or sometimes half a page is rendered and rest is just white, with no error, and a reload is needed to get the full page.

Customers are also randomly seeing roundcube webmail saying access to / is forbidden after clicking on a email. I'm not sure if these are related.

/var/log/httpd/error_log shows lots of these - 36 since 04:15 am (random times through the day, not terribly frequent and no obvious pattern to them) :

[Sun Jul 17 08:39:43.674098 2022] [core:error] [pid 23340:tid 140057360475904] (24)Too many open files: AH02179: apr_socket_accept: (client socket)
...
[Sun Jul 17 12:40:02.939600 2022] [core:notice] [pid 24439:tid 140058938951808] AH00052: child pid 30791 exit signal Segmentation fault (11)
[Sun Jul 17 12:55:05.836114 2022] [core:notice] [pid 24439:tid 140058938951808] AH00052: child pid 731 exit signal Segmentation fault (11)

Where can I start debugging these please?

24439 is /usr/sbin/httpd -DFOREGROUND and there are 483 child processes of this running, which I think is ok as I have quite a few websites running.

Thanks

 

[finbarr69]

Answering my own question...

https://support.plesk.com/hc/en-us/articles/213366549-How-to-enable-core-dumps-for-Apache-and-trace-Apache-segmentation-fault-on-a-Linux-server