HA Backup & Restore

We are using HA successfully on a two-server cluster. We are installing FreePBX on two new servers, and are migrating our data by using a backup taken from the primary server, freepbx-a.

When we restore the backup (which works fine, no errors), the HA functionality stops working. The reason appears to be that restoring the backup has changed the internal interface IP eth1 to what it was on the old server. We have tried manually setting the internal interface to the original randomly selected internal IP, and this did not make a difference.

When we run HA health check, we get the following errors:

We run the recommended commands (after changing rm -f to rm -rf. Please note that error in the instructions) but our secondary server is still NodeDown/Unknown

What parts of the backup should we select such that it will restore our trunking, extensions etc. but not break our working HA configuration on the new servers?

What is the correct procedure for rebuilding a HA cluster from backup? Were this a DR scenario, we would have lost HA functionality.

For Schmoozecom folks: We opened an official support ticket for this issue (252993) last Friday and have not had any replies, not even for assignment of the case. Can we please get some help on this issue ASAP?

ticket has been updated and i also left you a voicemail

please understand you opened the ticket with an account that does not have an SLA agreement - if you need an guaranteed response time on tickets please consider adding a POMPS agreement to the deployment - http://literature.schmoozecom.com/EUA/FreePBX-POMPs-EULA.pdf

I’ve created a ticket for this. This is a feature request - we’ve never really supported restoring BETWEEN clusters, as it’s tricky, but it’s something that should be doable.

http://issues.freepbx.org/browse/FREEPBX-8658

–Rob

Rob, thanks. We have modified our backup to exclude the freepbxha table and that resolves the problems we were seeing.

Another couple of questions I had:

We noticed certain services we do not use (namely isymphony) are started upon fail-over. Are services automatically restarted when they are in /drbd, and what’s the recommended way to disable them?

Conversely if we do want a service to failover between nodes does it need to be in /drbd?