| SYSTEM INFORMATION ||
| OS type and version | CentOS Linux 7.9.2009 |
| Webmin version | 1.994 |
| Virtualmin version | 7.1 |
| Related packages | Apache 2.4.6, PHP 8.1.9, PHP-FPM 8.1.9 Server |
When I set my sites to PHP-FPM all is just fine until after a server reboot. All processes are up & running (apache, PHP-FPM Server) but the websites throw an error 503 Service unavailable.
In the apache error log it shows
[proxy:error] [pid 1401] (2)No such file or directory: AH02454: FCGI: attempt to connect to Unix domain socket /var/php-fpm/160415238918520.sock (*) failed
[proxy_fcgi:error] [pid 1401] [client ip:63564] AH01079: failed to make connection to backend: httpd-UDS, referer: https://my.website.com/page
If I restart the PHP-FPM Server or if I leave it as is and change one of the sites back to FastCGI (or any of the other options) and immediately back to PHP-FPM it’s ok again (I guess that also triggers a restart of the PHP-FPM server process?)
It seems that after a reboot the directory /var/php-fpm is empty. After restarting PHP-FPM Server process it’s filled with the missing .sock files
This seems identical to the issue reported here: New sites php-fpm not working! but also after changing the PHP-FPM Server connection from socket files to TCP Port the behavior remains exactly the same. In the error log it shows
[proxy:error] [pid 1418] (111)Connection refused: AH00957: FCGI: attempt to connect to 127.0.0.1:8002 (*) failed
[proxy_fcgi:error] [pid 1418] [client my ip:64477] AH01079: failed to make connection to backend: localhost, referer: https://my.website.com/somepage
again the /var/php-fpm directory is empty (not sure if this is relevant with the TCP Port option) and after restarting the PHP-FPM server it’s all good again.
I have a test-server and a production server (same OS & Virtual/webmin versions) who both experience this same issue.
Please advise on how best to resolve this (as I workaround I guess I could add a bootup-script to restart the PHP-FPM server but I’d rather have a structural solution)