Eliminate Downtime with the New 3CX SBC High Availability Clustering - Technical Preview

[cobaltit]

@mariosM_3CX

Just checking on how things are looking on this front? Is this expected to be be stable with the release of Update 5

 

[ivank]

@mariosM_3CX
Just checking on how things are looking on this front? Is this expected to be be stable with the release of Update 5

Yes, it should be stable enough.
As to your original request, we need to understand if it was a cluster-related problem, or something else.

If according to crmsh cluster was in good shape (i.e., both nodes agree on who's active and active is showing all resources in green), that means cluster itself has nothing to do with the problem.
Further investigation should be done in 3cxsbc.log to find the reason why it can't connect to the PBX.

If on the other hand you see that 3cxsbc resource cannot start properly and keeps restarting, it's a completely different story. Deep understanding of pacemaker is needed.

At least, you could describe how it looked from PBX point of view and what you've eventually done to resolve the situation.

 

[cobaltit]

@ivank

So that was a few weeks ago so I don't remember entirely but per crm_mon both instances were up and agreed on who was active. I want to say the SBC/trunk was showing green in 3CX but I don't remember. Rebooting the SBCs individually fixed the problem. I'll see if the 3cxsbc log is still around from then.

 

[jparfrey]

"be able to send and receive ICMP (ping) messages to and from the PBX host"
To be sure, is the PBX pinging the RPI to check live status, if so how do you handle this behind a NAT router?

 

[ivank]

To be sure, is the PBX pinging the RPI to check live status, if so how do you handle this behind a NAT router?

Each SBC node pings the PBX.

 

[accentlogic]

Each SBC node pings the PBX.

Good to know. AWS by default does not allow ICMP (ping) and users will need to enable it if using that platform - or any other that disables it by default. Amazon LightSail currently does not allow ping at all, so users on that platform will need to move it if they want to use SBC clustering.

 

[VoIP-Cloud]

Amazon LightSail currently does not allow ping at all, so users on that platform will need to move it if they want to use SBC clustering.

well that is not the full story. Amazon Lightsail sure does allow icmp. But you can't define / allow it in their firewall, so if you want or need icmp you need to use IPtables on the 3CX and SBC clustering will work with Amazon Lightsail.

Edit: was wrong, too late, wrong cloud - my fault you are right, LightSail still doesn't allow icmp

 

[VoIP-Cloud]

referring to Lightsail, does anyone have problems with cpu steal time? As the cheaper instances are only 1 vCPU and shared. 2 vCPUs are $20 and more

 

[cobaltit]

referring to Lightsail, does anyone have problems with cpu steal time? As the cheaper instances are only 1 vCPU and shared. 2 vCPUs are $20 and more

I'm sure people who are overloading it are having CPU problems. I mean that's like asking if people are having disk space problems. People doing a lot of recording will have disk space problems, people who aren't won't. If you are having high CPU load then you'll need to upgrade your instance.

https://www.3cx.com/docs/recommended-hardware-specifications-for-3cx/