Manual failover

[CloudAware]

Hello everybody,

I am looking at configuring a 3CX failover configuration. The failover is going to be done in DNS, so no SBC. I am running an Enterprise license.
I have the following questions:
- I really would like to do manual failovers. This is to perform maintenance during working hours. I do understand that removes the redundancy for the time i am doing maintenance, but it will also provide me with more normal working hours. I see that there is no real way to do manual failover other than stopping services, but when updating/restarting the server under maintenance it may show up as available again for the failover server and it will try to switch back. Given the TTL of DNS (5M) this is undesireable. Can i do something like a "manual maintenance failover"?
- I am a bit concerned about lack of quorum and the dreaded splitbrain when implementing the failover. I would rather prefer to have two identical servers up and running, both with working trunks and externally manually switch DNSses. Is this possible with an Enterprise license (not technically, but license-wise)? Can i run two active servers (with only one having phones registered) on a single enterprise license?

Thank you all for your insight!

 

[Evolute IT]

Failover is supported on Pro and Enterprise editions. On Enterprise, you indeed get a standby license for the passive server. However, this will always be a Active/Passive configuration.

Also, if using the 3CX FQDN, the DNS will be automatically updated. So, no need to mess around with that.

Configuring it is fairly simple when you follow the docs: https://www.3cx.com/docs/failover/

 

[CloudAware]

Thank you Frederick,

Unfortunately my questions are a result of (successfully) configuring failover using an enterprise 3CX license. Do you have any insights in these questions?
Quorum is very important to use so using the automatic failover in 3CX is useless to us. We are forced to implement this ourselves. Hence my 2 questions above.
Thank you very much!

 

[Evolute IT]

Quorum is very important to use so using the automatic failover in 3CX is useless to us. We are forced to implement this ourselves. Hence my 2 questions above.

What is Quorum? And no, there's no way to do your own failover unless you use VMware vMotion or stuff like that which is more HA than failover.

 

[CloudAware]

So i still have my two questions actually.
Short versions:
1) Can i do the failover manually?
2) Does the Enterprise license allow me to run two instances where on only one instance extensions are registred?

Thank you very much!

 

[CloudAware]

For archive purposes:

We have had contact with the NL distributor of 3CX. Although they acknowledge the issues with failover they indicate that a failover will require a manual action to fall back to the primary server. Although i think this is still a bit of an ugly hack, it prevents servers from flapping back and forth. It also gives us *sort of* a way to manually fail over if we want to do maintenance on the prim. server.

Concerning the license: they have confirmed that in the enterprise license you can have two servers up and running with all services running in a single licence (using the same FQDN).

So the solution we have chosen is:
- DISable failover
- ENable automatic backup/restore between the two servers
- Potentially do a manual backup and restore between the servers if we are doing a lot of work during the day.
- Solely use (self hosted) DNS to fail over (because of outage or maintenace)

I hope this helps others who have read the failover docs and are still struggling with the ins and outs of how it works.

 

[ChrisC_3CX]

Just want to add that we do not recommend having two 3CX Instances, using the same License Key partially or fully up and running unless in a 3CX Failover configuration. Issues could arise regarding activation, SIP Trunk Registration and Incoming/Outgoing call routing as well as 3CX FQDN resolution to the correct IP.

 

[NickD_3CX]

For archive purposes:

We have had contact with the NL distributor of 3CX. Although they acknowledge the issues with failover they indicate that a failover will require a manual action to fall back to the primary server. Although i think this is still a bit of an ugly hack, it prevents servers from flapping back and forth. It also gives us *sort of* a way to manually fail over if we want to do maintenance on the prim. server.

Concerning the license: they have confirmed that in the enterprise license you can have two servers up and running with all services running in a single licence (using the same FQDN).

So the solution we have chosen is:
- DISable failover
- ENable automatic backup/restore between the two servers
- Potentially do a manual backup and restore between the servers if we are doing a lot of work during the day.
- Solely use (self hosted) DNS to fail over (because of outage or maintenace)

I hope this helps others who have read the failover docs and are still struggling with the ins and outs of how it works.

Hi!

Just one note on this, if you Active and Passive servers are on different Public IPs, then bringing both to the "Active" state would case a "flapping" of the WAN IP the FQDN would resolve to (assuming you're using a 3CX FQDN).
If they are both behind the same Public IP, then this shouldn't be an issue.

 

[CloudAware]

Just want to add that we do not recommend having two 3CX Instances, using the same License Key partially or fully up and running unless in a 3CX Failover configuration. Issues could arise regarding activation, SIP Trunk Registration and Incoming/Outgoing call routing as well as 3CX FQDN resolution to the correct IP.

Thank you. This is a useful reply. I might add though that we have our own voip infrastructure, so we can dynamically route DID´s to the proper PBX. I do understand the 3CX FQDN DNS challenges involved, but as stated, we are using our own DNS infrastructure, so this is not an issue to us.

 

[CloudAware]

Hi!

Just one note on this, if you Active and Passive servers are on different Public IPs, then bringing both to the "Active" state would case a "flapping" of the WAN IP the FQDN would resolve to (assuming you're using a 3CX FQDN).
If they are both behind the same Public IP, then this shouldn't be an issue.

I guess we solved this by not using the failover functionality and therefore not letting 3CX control the DNS.

 

[ChrisC_3CX]

I completely understand, I just mentioned it in case there are others reading this that do not have a similar setup. Since you are managing your own DNS that part of it at least should not be an issue.