Hello, I’m experiencing a strange intermittent issue on a cPanel server running CloudLinux, where some domains randomly hang during page loads without returning any error. Usually they partially load (I.e. only html without css/js). Website normally loads upon refresh and during the next 20-30 page views the same problem happens again.
The environment uses Apache 2.4 with MPM event and PHP via CGI (PHP-FPM is disabled). The issue occurs across PHP 7.4, 8.0 and 8.2. HTTP/2 has been tested both enabled and disabled, and CageFS and ModSecurity were also temporarily disabled for testing, with no real change in behavior.
No 500 or 403 errors are logged, access logs continue normally, error logs are clean, and there are no Apache crashes or coredumps. The problem is reported by users from multiple ISPs and does not correlate with high server load.
At the system level, there appears to be an increased number of TCP retransmissions when the issue occurs.
Has anyone encountered similar behavior with Apache and CloudLinux where connections appear to stall without visible errors? Any insight or direction would be greatly appreciated.
The environment uses Apache 2.4 with MPM event and PHP via CGI (PHP-FPM is disabled). The issue occurs across PHP 7.4, 8.0 and 8.2. HTTP/2 has been tested both enabled and disabled, and CageFS and ModSecurity were also temporarily disabled for testing, with no real change in behavior.
No 500 or 403 errors are logged, access logs continue normally, error logs are clean, and there are no Apache crashes or coredumps. The problem is reported by users from multiple ISPs and does not correlate with high server load.
At the system level, there appears to be an increased number of TCP retransmissions when the issue occurs.
Has anyone encountered similar behavior with Apache and CloudLinux where connections appear to stall without visible errors? Any insight or direction would be greatly appreciated.
I would move off of the prefork mpm to event and move from cgi to lsapi for the PHP handler
That said - as you can reproduce this …. strace the processes and see what they’re doing or hire an experienced and competent systems administrator to assist you.
That said - as you can reproduce this …. strace the processes and see what they’re doing or hire an experienced and competent systems administrator to assist you.
When it’s happening, I’d check server-status (just enable it temporarily and lock it down to your IP).
If you see a lot of workers sitting in W / R / K states for the affected vhosts (sending reply, reading request, keepalive), then Apache itself is likely where things are getting stuck.
If the workers look normal but clients are still hanging, I’d start looking more at network-side issues (NIC, firewall, conntrack, etc.) rather than Apache.
I’d also run:
apachectl fullstatus
(or check Apache Status from WHM under Service Status)
If you see a lot of workers sitting in W / R / K states for the affected vhosts (sending reply, reading request, keepalive), then Apache itself is likely where things are getting stuck.
If the workers look normal but clients are still hanging, I’d start looking more at network-side issues (NIC, firewall, conntrack, etc.) rather than Apache.
I’d also run:
apachectl fullstatus
(or check Apache Status from WHM under Service Status)
Are there any similarities between the websites the issue is happening on? Sounds like something might be trying to call home (a licensing script?) but the connection is timing out.
Because of development reasons. I am using this also in other servers without problem.
Development of what?
It seems you are using the default settings from EA4 without optimizing them properly.
CGI is slow and insecure. Take some time to understand different PHP handlers, and you will see the difference in speed and security. The hanging issues will go away once you optimize it properly.
- Page :
- 1
- 2
There are no replies made for this post yet.