EasyApache install with php suexec caused all php to result in internal server errors

[Solokron]

Just updated server from PHP 4.4.2 with php suxec to 4.4.4 with php suexec. Any call to PHP results in an internal error. Again, these clients have been in a suexec environment for quite a while and the php upgrade was performed. Permissions and ownership are still what are necessary for php suexec. Even a simple php file with the contents "hello world" chowned and chmod 755 results in Internal Server Error. I tried thunking back to 4.4.2 and php suexec and this still occurs.

[AndyReed]

Just updated server from PHP 4.4.2 with php suxec to 4.4.4 with php suexec. Any call to PHP results in an internal error. Again, these clients have been in a suexec environment for quite a while and the php upgrade was performed. Permissions and ownership are still what are necessary for php suexec. Even a simple php file with the contents "hello world" chowned and chmod 755 results in Internal Server Error. I tried thunking back to 4.4.2 and php suexec and this still occurs.

Upgrade ZendOptimizer to the latest release v3.0.1, if using ZendOptimizer. If not, re-compile Php with Verbose on and look for errors.

[Solokron]

3.01 was already installed. I did in fact reinstall it afterwards to no resolve.

When I run easyapache I run from an ssh session.

Upgrade ZendOptimizer to the latest release v3.0.1, if using ZendOptimizer. If not, re-compile Php with Verbose on and look for errors.

[WebScHoLaR]

Have you checked /usr/local/apache/logs/error_log and /usr/local/apache/logs/suexec_log for the errors?

[Solokron]

Nice FF avatar. Yes I checked suexec immediately when it started occuring but could not locate any errors:

[2006-06-28 15:29:35]: info: (target/actual) uid: (persocom/persocom) gid: (persocom/persocom) cmd: home.php
[2006-06-28 15:29:35]: info: (target/actual) uid: (persocom/persocom) gid: (persocom/persocom) cmd: adlog.php
[2006-06-28 15:29:35]: info: (target/actual) uid: (persocom/persocom) gid: (persocom/persocom) cmd: adlog.php
[2006-06-28 15:29:35]: info: (target/actual) uid: (persocom/persocom) gid: (persocom/persocom) cmd: adlog.php
[2006-06-28 15:29:36]: info: (target/actual) uid: (persocom/persocom) gid: (persocom/persocom) cmd: adlog.php
[2006-06-28 15:29:36]: info: (target/actual) uid: (persocom/persocom) gid: (persocom/persocom) cmd: adlog.php
[2006-06-28 15:29:36]: info: (target/actual) uid: (persocom/persocom) gid: (persocom/persocom) cmd: adlog.php

Which is to be expected. Even simple hello files with .php for an extension were hosed.

error_log:

[Fri Aug 25 15:08:58 2006] [error] [client 67.160.188.165] Premature end of script headers: /home/persocom/public_html/online_dating.php
failed to open log file
fopen: Permission denied

[Fri Aug 25 16:48:33 2006] [error] [client 72.30.111.152] Premature end of script headers: /home/persocom/public_html/ad_server/

These files were indeed set to 755 and chowned correctly which threw me off.


I tried recompliling back to apache with suexec and php 4.4.2 but still ocurs so I have recompiled apache without suexec on the server to get it back up and running which has not made me too happy.

I will try again tonight with a release build and httpd configtest

Have you checked /usr/local/apache/logs/error_log and /usr/local/apache/logs/suexec_log for the errors?

[Solokron]

Well here we go. It appears easy apache, even with a stable build and upcp --force selected does what the beta apparently does which I ran into in a thread researching this issue. It rems out

LoadModule php4_module libexec/libphp4.so
AddModule mod_php4.c

By removing the rems and restarting Apache it loads back up with php suexec once again. Or am I making the mistake of simply enabling the standard php4 module and phpsuexec is being bypassed?

[Solokron]

If I remark back, 500 internal errors again.

error_log:

[Tue Aug 29 04:41:49 2006] [error] [client 71.231.237.36] Premature end of script headers: /home/userxxx/public_html/index.php
[Tue Aug 29 04:41:49 2006] [error] [client 71.231.237.36] File does not exist: /home/userxxx/public_html/500.shtml

suexec_log
[2006-08-29 04:41:49]: info: (target/actual) uid: (userxxx/userxxx) gid: (userxxx/userxxx) cmd: index.php

[NoAgendas]

I get internal 500 errors on one particular server after going from php 4.4.2 to 4.4.4 (all modules the same except newer php version)

apache log:

Premature end of script headers

(all php based sites)

What is the fix for this?

Another problem with fastcgi:

/etc/rc.d/init.d/httpd restart: httpd not running, trying to start
Syntax error on line 193 of /usr/local/apache/conf/httpd.conf:
FastCgiIpcDir /tmp/fcgi_ipc: access for server (uid -1, gid -1) failed: write not allowed

[dave9000]

Well here we go. It appears easy apache, even with a stable build and upcp --force selected does what the beta apparently does which I ran into in a thread researching this issue. It rems out

LoadModule php4_module libexec/libphp4.so
AddModule mod_php4.c

By removing the rems and restarting Apache it loads back up with php suexec once again. Or am I making the mistake of simply enabling the standard php4 module and phpsuexec is being bypassed?

Your bypassing the phpsuexec when you re-enable the loadmodule and addmodule lines

what has worked for me with the fopen log error is delete the log files , restart the apache and let it remake them

also check and make sure /tmp is not full and has these permissions

/tmp 1777 root:root

[NoAgendas]

I get internal 500 errors on one particular server after going from php 4.4.2 to 4.4.4 (all modules the same except newer php version)

apache log:

Premature end of script headers

(all php based sites)

What is the fix for this?

Another problem with fastcgi:

/etc/rc.d/init.d/httpd restart: httpd not running, trying to start
Syntax error on line 193 of /usr/local/apache/conf/httpd.conf:
FastCgiIpcDir /tmp/fcgi_ipc: access for server (uid -1, gid -1) failed: write not allowed

How about my problem here?

[dave9000]

I get internal 500 errors on one particular server after going from php 4.4.2 to 4.4.4 (all modules the same except newer php version)

apache log:

Premature end of script headers

(all php based sites)

What is the fix for this?

Another problem with fastcgi:

/etc/rc.d/init.d/httpd restart: httpd not running, trying to start
Syntax error on line 193 of /usr/local/apache/conf/httpd.conf:
FastCgiIpcDir /tmp/fcgi_ipc: access for server (uid -1, gid -1) failed: write not allowed

check the permissions on the /tmp/fcgi_ipc make sure the web server can write to that directory and that the directory exists

[NoAgendas]

I had done that already, what else would cause this?

[dave9000]

I have no experience with fastcgi so I am not going to be much additional help here

[NoAgendas]

drwxrwxrwx 3 nobody nobody 4.0K Jul 29 11:51 fcgi_ipc/

[Solokron]

I have performed these to no resolve.

Your bypassing the phpsuexec when you re-enable the loadmodule and addmodule lines

what has worked for me with the fopen log error is delete the log files , restart the apache and let it remake them

also check and make sure /tmp is not full and has these permissions

/tmp 1777 root:root