• If you are still using CentOS 7.9, it's time to convert to Alma 8 with the free centos2alma tool by Plesk or Plesk Migrator. Please let us know your experiences or concerns in this thread:
    CentOS2Alma discussion

major issue in fastcgi, plesk 9.3

J

joshua strebel

Guest
I cannot even isolate the cause. I have over 100 vhosts running, I had them all set to run fastcgi.

The server is a massive 8core xeon, 8gbs ram.

I think there is a bug in the fastcgi wrapper.. inside of plesk.

Error logs would fill up with the following error, and then apache would just quit responding period. Would require a reboot of the entire vps. Then after 5-10 minutes same thing. I mass edited the domains and flipped them all to php-cgi mode, seems to be okay now.

Errors:

[Thu Jan 28 10:40:35 2010] [warn] (103)Software caused connection abort: mod_fcgid: ap_pass_brigade failed in handle_request function
[Thu Jan 28 10:40:38 2010] [warn] (103)Software caused connection abort: mod_fcgid: ap_pass_brigade failed in handle_request function
[Thu Jan 28 10:40:45 2010] [warn] (103)Software caused connection abort: mod_fcgid: ap_pass_brigade failed in handle_request function
[Thu Jan 28 10:40:48 2010] [warn] (103)Software caused connection abort: mod_fcgid: ap_pass_brigade failed in handle_request function

Server resources would not even blip or spike.. apache would just freeze solid.
centos5, plesk 9.3, php 5.2.12, apache 2.2.*

Any ideas?
 
Try to google "Software caused connection abort: mod_fcgid: ap_pass_brigade failed in handle_request function"
There are a lot of mentions about this problem.
 
So just for testing sake i flipped 100 domains from php_cgi to fast_cgi and of course.. random timeouts all over the place again. Plesk's support of fast_cgi is pretty weak
 
Aint no bug in Plesk - they're going to be using stock packages from your OS provider.

Just FYI, we have no problems with PHP as FCGI on Debian with many more than 100 virtual hosts (all of our sites run PHP as FCGI).

Make sure you're tuning the FCGI configuration too, to your environment. I'd say what you're experiencing is just time out related. Make sure too you're not running an opcode cache w/ PHP/FCGI too, as each virtual host will end up with its own cache chewing in to your RAM.
 
Aint no bug in Plesk - they're going to be using stock packages from your OS provider.

Just FYI, we have no problems with PHP as FCGI on Debian with many more than 100 virtual hosts (all of our sites run PHP as FCGI).

Make sure you're tuning the FCGI configuration too, to your environment. I'd say what you're experiencing is just time out related. Make sure too you're not running an opcode cache w/ PHP/FCGI too, as each virtual host will end up with its own cache chewing in to your RAM.

For me these don't look like stock OS packages (CentOS 5.4 , Plesk 9.3):

# rpm -qa | grep cgi
psa-mod_fcgid-1.10-3
psa-fcgi-2.4.0-2
psa-mod-fcgid-configurator-1.0-14

They look like packages distributed by parallels?

This problem can be replicated with a small PHP script just calling "sleep 60;"

Normally in the vhost.conf file these commands would help this sort of error, but, doesn't for this one:

ProcessLifeTime 7200
IPCCommTimeout 7200
IPCConnectTimeout 300

Still looking for a solution for this.

jacko
 
Back
Top