4

I am migrating mail servers from OVH/Centos7 to AWS/Amazon Linux; part of my setup uses roundcube webmail.

I have copied all of my config and supporting files from one instance to the other, however can not seem to make the webmail app load as expected. Anything and everything produces a 500 with no useful errors displayed anywhere.

All apps and subdirectories are owned by apache:apache

With the following error config in php.ini

display_errors = On
error_reporting = E_ALL
log_errors = On

This is the ONLY log entry created when navigating to my URL:

w.x.y.z - - [27/Dec/2018:03:56:03 +0000] "GET / HTTP/1.1" 500 - "-" "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/71.0.3578.98 Safari/537.36"

The log entry clearly shows the 500, but produces no errors. In addition, the error log file within the roundcube directory is also empty.

The response to the client is empty, just a 500.

$curl https://mail.example.com/

Header is 500:

$curl https://mail.example.com/ --head
HTTP/1.1 500 Internal Server Error
....

If I cd to the roundcube directory and try to evaluate the index, it returns nothing:

$cd /path/to/roundcube
$php index.php

$

I have tried to disable every single roundcube plugin, in config/config.inc.php

$config['plugins'] = array();

Every single other app and web page works, after the transplant; roundcube is the only thing which is throwing 500's and not producing any error messages...

syslog is empty of relevant messages.

What can I possibly do to further investigate why this might be failing?

Matt Clark
  • 27,671
  • 19
  • 68
  • 123
  • If the specific vhost entry for this app doesn't have an `ErrorLog` set, try checking the default Apache error log – Phil Dec 27 '18 at 04:17
  • 1
    I do actually have custom logs set, `CustomLog` and `ErrorLog` both defined. Custom log shows shows the above 500 entry, and the ErrorLog is empty.. – Matt Clark Dec 27 '18 at 04:19
  • 1
    Since you have `log_errors = On`, where is PHP logging them? Forgive the noob question but have you reloaded / restarted Apache after making any changes to config (just covering all bases)? – Phil Dec 27 '18 at 04:22
  • That is what I am wondering myself, and the reason for the question... I have absolutely restarted the process (and the server) I've checked the httpd general logs, outside of my defined VHost logs, syslogs, and roundcube logs. I have no idea where to see the actual errors are being logged. If I create a php file in the same directory, I can view it and see errors generated. Permissions all the same. SELinux is not enabled. – Matt Clark Dec 27 '18 at 04:25
  • ughhhhhhh..... roundcube itself has an `.htaccess` file which sets: `php_flag display_errors Off` – Matt Clark Dec 27 '18 at 04:43
  • 1
    Set the `error_log` in `php.ini`, and make sure your webmail app isn't overriding your error directives. Do some debugging on the webmail app, throw some `var_dump()` statements into it. – miken32 Dec 27 '18 at 04:53
  • does your apache process user have write privilege for wherever some logging will take place ? – YvesLeBorg Dec 27 '18 at 08:52

0 Answers0