fcgi unix domain socket/TCP socket question

["Rose, John B" <jbrose@xxxxxxxxxxxxxxxxxxx>]

Couple of questions ...

Reading this Apache httpd document ...

https://httpd.apache.org/docs/2.4/mod/mod_proxy_fcgi.html

Does the below mean if the max children for a PHP-FPM pool using Unix Domain Sockets is reached then any subsequent PHP-FPM requests for that virtual host utilize a TCP socket? If we have the base www.conf pool configured to use TCP as opposed to Unix Domain Sockets?

SetHandler "proxy:unix:/var/run/php-fpm/something.com.sock|fcgi://something.com/"

What would be the impact of NOT defining a matching worker for the above config? Like is done in this line on the apache.org documentation page referenced above ..

<Proxy "fcgi://localhost/" enablereuse=on max=10>

Occasionally, typically when someone is doing a Wordpress core or plugin upgrade, all PHP-FPM sites cease to load their content. And a reboot is needed to get back to normal. But a large number of PHP-FPM processes get started

seemingly maxing out the "Sethandler" connections. then we see "fcgi" timeouts in log.

Our goal is to avoid the perceived problem the results from this php-fpm/TCP bug by using Domain Sockets. Any suggestions?

https://blog.cloudflare.com/this-is-strictly-a-violation-of-the-tcp-specification/

Thanks for everyones work on Apache