Having to manually restart httpd for addon domain

[Tam]

Looking for pointers ont his one ...

each time an addon domain is added via a cPanel account it is showing the default Apache Success page, a manual httpd restart solves this (the site shows up correctly after that) but this is obviously not desirable.

Any suggestions gratefully received.

thanks in advance.

[Infopro]

cPanel up to date?

[Tam]

Yes, /scripts/upcp --force has been run.

[thewebhosting]

Please check the CPanel error logs if you find anything there.

The path is /usr/local/cpanel/logs/error_logs

[Tam]

4 repetitve errors ...

Mon Dec 8 11:23:22 2008 info [bin/parkadmin.pl]: Global NS entry #2 is not a valid domain
Mon Dec 8 11:23:22 2008 info [bin/parkadmin.pl]: Global NS entry #3 is not a valid domain
Mon Dec 8 11:23:22 2008 info [bin/parkadmin.pl]: reseller NS entry #2 is not a valid domain
Mon Dec 8 11:23:22 2008 info [bin/parkadmin.pl]: reseller NS entry #3 is not a valid domain

the domain is of-course valid but is hosted on another server and the DNS zone on this server reflects precisely what is on the other server.

[Tam]

Given that this occurs with any reseller/user account I doubt it is not nameserver related. Instead it could be apache related - from the apache error log when adding an addon domain ...

[Mon Dec 08 14:47:42 2008] [notice] Graceful restart requested, doing restart
[Mon Dec 08 14:47:42 2008] [error] (9)Bad file descriptor: apr_socket_accept: (client socket)
[Mon Dec 08 14:47:42 2008] [error] (9)Bad file descriptor: apr_socket_accept: (client socket)
[Mon Dec 08 14:47:42 2008] [error] (9)Bad file descriptor: apr_socket_accept: (client socket)
[Mon Dec 08 14:47:42 2008] [error] (9)Bad file descriptor: apr_socket_accept: (client socket)
[Mon Dec 08 14:47:42 2008] [error] (9)Bad file descriptor: apr_socket_accept: (client socket)
[Mon Dec 08 14:47:42 2008] [error] (9)Bad file descriptor: apr_socket_accept: (client socket)
[Mon Dec 08 14:47:42 2008] [error] (9)Bad file descriptor: apr_socket_accept: (client socket)
[Mon Dec 08 14:47:42 2008] [error] (9)Bad file descriptor: apr_socket_accept: (client socket)
[Mon Dec 08 14:47:42 2008] [error] (9)Bad file descriptor: apr_socket_accept: (client socket)
[Mon Dec 08 14:47:42 2008] [error] (9)Bad file descriptor: apr_socket_accept: (client socket)
[Mon Dec 08 14:47:46 2008] [notice] mod_bw : Memory Allocated 0 bytes (each conf takes 28 bytes)
[Mon Dec 08 14:47:46 2008] [notice] mod_bw : Version 0.8 - Initialized [0 Confs]
[Mon Dec 08 14:47:46 2008] [notice] Apache/2.2.10 (Unix) mod_ssl/2.2.10 OpenSSL/0.9.8b mod_auth_passthrough/2.1 mod_bwlimited/1.4 FrontPage/5.0.2.2635 mod_perl/2.0.4 Perl/v5.8.8 configured -- resuming normal operations

[Server-Rack]

This is probably not the solution you were looking for but, you may consider getting a reseller account at the web host of your choice. Pricing is very competitive now and your site may do better standing on its own. This may uncomplicate things for you.

- Rick

[Tam]

This is probably not the solution you were looking for but, you may consider getting a reseller account at the web host of your choice. Pricing is very competitive now and your site may do better standing on its own. This may uncomplicate things for you.

- Rick

We've been supplying reseller accounts for years, and this is occuring on a dedicated server of ours.

[Tam]

Seems to be a bug with Apache MPM prefork and graceful restarts according to https://issues.apache.org/bugzilla/s...g.cgi?id=42829

Can cpanel confirm if the patch is/will be included in easyapache?

I'm going to try rebuilding apache without prefork to see if it is still an issue.

[cpanelkenneth]

Seems to be a bug with Apache MPM prefork and graceful restarts according to https://issues.apache.org/bugzilla/s...g.cgi?id=42829

Can cpanel confirm if the patch is/will be included in easyapache?

I'm going to try rebuilding apache without prefork to see if it is still an issue.

A patch will only be commited to EasyApache once the Apache project is satisfied with the quality of the patches. At this time the best work around is to use Apache 2.0.

Some information about this is available at http://www.cpanel.net/support/docs/e..._g_status.html

[Tam]

Thanks very much Kenneth, much appreciated.

Removing prefork didn't resolve it anyway and I've also read somewhere else that mod_ssl can cause the issue but we cannot do without that. Anyway, Apache 2.0 doesn't have the issue, thanks very much for that.

[Tam]

grrr, rolling back to apache 2.0.63 did solve the issue last night but now the same problem returns, no error showing in the apache error log though

[rking]

I, too, am having this same issue.

I am not experiencing the Apache errors that Tam mentioned, but I do see this in my cPanel error_log:

2009-01-22 23:03:00 info [parkadmin] Global NS entry #2 is not a valid domain
2009-01-22 23:03:00 info [parkadmin] Global NS entry #3 is not a valid domain

I don't believe this is necessarily related to the Apache bug you've referenced. We have been running Apache/2.0.63 for months now. It has just come to my attention today that several users of ours were contacting us every other week or so asking for an Apache restart after they add their domains. Different techs were receiving the requests and didn't realize it was a recurring issue.

This issue seems to be global to all users, though I have so far only tested 2 users specifically.

[rking]

Apache doesn't seem to add the domain and then restart in the correct order. We don't have a problem with Apache restarting or the proper entries being added, so we have a suggestion for a temporary fix.

Clients can add the domain they want to add first. Apache will restart and the entry will be in the httpd.conf but it will not be in the running config. Then, add a fake domain. Apache will be restarted with the first domain now in effect, but the fake domain won't. Remove the fake domain.

Still inconvenient, but not as inconvenient as having to contact us for restarts.

I'll be running a forced cPanel update to see if it corrects this issue and then filing a ticket with cPanel. Updates to follow.

[rking]

No love. Forced cPanel update didn't fix it. Submitting a ticket.