FreePBX HA-Apache wont start on second server

Problem: httpd_service won't start on freepbx-

httpd_service was discovered to be unknown error on freepbx-b

It is possible for the installer on a new system to not complete the migration of httpd on the second server. You can check this by ssh-ing to freepbx-b and checking the /etc/httpd directory.

-bash-4.1$ hostname freepbx-b -bash-4.1$ ls -ald /etc/httpd drwxr-xr-x. 4 root root 4096 Nov 10 10:47 /etc/httpd

Notice how this is NOT a symlink. A correctly migrated node should look like this:

-bash-4.1$ ls -ald /etc/httpd lrwxrwxrwx 1 root root 15 Nov 12 13:51 /etc/httpd -> /drbd/httpd/etc -bash-4.1$

(Note that it will be red if that service is not running on that node. This is to be expected.)

Resolution

On freepbx-b, run these commands as the root user:

cd /var/www/html/admin/modules/freepbx_ha ./moveapache.php

You will see several screens of WARNINGs, but after that's completed, checking the /etc/httpd directory will then show the symlink as expected. 

To ensure the cluster is aware that this problem is fixed, reboot the freepbx-b node.

Problem: Services aren't starting, no errors displayed

This can happen when testing manual failover (eg, running 'asterisk -rx core stop now' several times) to simulate service failure.

It's possible for the visible (displayed on the GUI) and internal (accessible only to pacemaker) error counters to get out of synchronization. This causes services to mark themselves as 'not startable' when there is no visible indication of this.

Resolution

Make sure that the other node is not in standby, and reboot the node that is refusing to start services.

This error will self-heal, but this type of issue would not occur if you have fencing set up. 

Return to Documentation Home I Return to Sangoma Support