Split DNS with Failover scenario

[DXC]

Hi,

We are using failover scenario (active/passive server).

One of the several requirements before upgrade to V20 is Split DNS.
How should FQDN be resolved in this case?

When the passive server becomes the active one, an IP address is changing.

 

[Alejandro_3CX]

Hi if you are using a failover with both PBX in the same LAN, then it has been mandatory for a long time to use an internal FQDN, it means you have a managed network with an internal DNS server where you created an A Record for this internal FQDN that resolves the local IP of the active server. So when the failover occurs, you should have uploaded a script in the passive server to update this A Record in the DNS Server, so the internal FQDN starts resolving the new local IP address.

For IP phones in the same LAN, these should have been provisioned pointing to this internal fqdn, so it will bee transparent for these devices.

 

[DXC]

My question is related to a public IP address.

Two servers are situated in different locations and they have different public IP's.

 

[Alejandro_3CX]

As shown in the document If both pbx are in the cloud the external FQDN will resolve the public IP address of the active server.

Once the failover occurs the other server must be deactivated with a script, otherwise if it takes the role of active again, it will produce a conflict with the public IP address and the fqdn.

 

[DXC]

the other server must be deactivated with a script

What do you mean?

 

[Alejandro_3CX]

Ok, lets keep on the failover scenario installed both in the cloud as mentioned in the guide.

FQDN resolves the public IP of the Active Server. When there is a failover, the server that was as passive takes the active role, so the FQDN now resolves the new public IP address. Once the failover occurs, the server that failed must be shut down, if you leave it as it is without any intervention and the services are recovered, then you will have an issue with the fqdn resolving another IP. That is what I meant.

 

[DXC]

Got it. Thank you.

Unfortunately, 3CX has designed the failover scenario using a very bad concept, and as far as I know 3CX doesn't have a plan to improve the current behavior.

What we have for now:
- One way check. When the active server is recovered, no any automatic actions perform. We need to do it manually.
- We need to shut down the failed server manually immediately.

 

[Alejandro_3CX]

When the active server is recovered, no any automatic actions perform. We need to do it manually

Yes, you have to configure this server as passive now, no need to set is as passive again, otherwise you will be creating a new disruption.

- We need to shut down the failed server manually immediately.

or upload a script in the passive server.

 

[DXC]

Yes, you have to configure this server as passive now, no need to set is as passive again, otherwise you will be creating a new disruption.

Let me remind you that in such case we loose all previous statistics and that's a big problem for creating the reports.

 

[Alejandro_3CX]

of course, since the backup is restored once in the day and not every minute or hour, if the call log is very important to run the reports you may use the CDR instead so you get all the data in a different server.

The failover works as it is