Failover best practice and how do you return to the previous state

[Clint Reed]

Hello All,

We have a client that wanted a failover 3cx system v16.0.8.9. I have the 2 servers set up in AWS in different regions. The 2 sites were serviced by the 3cx system have their own SBC's, note there is VPN between the 2 sites, the phones are pointing at the sbc IP of the site. Now yesterday AWS had issues and the primary/active server went offline long enough for the passive to take over.

Questions -

1. How does the server switch back to the original state? Does anything need to be done or is it automatic?(I know now it's not as the primary server was back online) Whats the correct process to return to the original state?

2. Will this affect the SBC's, I do have the SBC failover IP set on the SBC's pointing to the IP of the passive server?

3. I also saw a post on issues with the backup/restore process if the server is down for a longer period of time. What is the best practice for that issue?

This is a standup-only build and I need it to be simple for the client to manage if the primary goes down.

One thought I had was to have them switch roles and let the secondary server become the primary by switching who is doing the backup and who doing the restores in the backup/restore area.

Thanks
Clint Reed
EasyIT

 

[Clint Reed]

Additionally, I know the documentation states to shut down the primary server if it fails, yesterdays case could not access it to turn it off as we had no access to the AWS console. What could be the negative result(s), other than that back and restore process likely lost their voicemails from yesterday.

 

[VasilisV_3CX]

Hi @Clint Reed,

To answer your questions:

1. How does the server switch back to the original state? Does anything need to be done or is it automatic?(I know now it's not as the primary server was back online) Whats the correct process to return to the original state?

This would happen manually. What you should be doing is bringing your active server back up, switching the passive server back to passive and then checking shortly afterwards that the FQDN is pointing to the correct external IP address.

2. Will this affect the SBC's, I do have the SBC failover IP set on the SBC's pointing to the IP of the passive server?

Yes, it will affect the SBCs but as the SBC is pointing to the FQDN, this shouldn't be a problem for too long, on ENT license keys the DNS propagation is 5 minutes. It's recommended that you use Google DNS settings (8.8.8.8) as your ISPs DNS may take longer to propagate.

3. I also saw a post on issues with the backup/restore process if the server is down for a longer period of time. What is the best practice for that issue?

The backup and restore functions should be set apart by a logical amount of time (if it's a small backup, and takes 5 minutes, set the restore to start 10 minutes after the backup etc).

What could be the negative result(s), other than that back and restore process likely lost their voicemails from yesterday.

You will lose any data between the time of the backup of the active, till the passive kicks in. So for example, if you set your backup to run at 00:01, and the passive kicks in at 03:00, you will have lost the data between those times. This will include voicemails as you've mentioned, as well as call logs, chat logs etc.