SIP HA IP address takeover - Linux on AWS

[gen_den]

Hi All,

Has anyone configured the Genesys SIP HA IP address takeover solution on Linux on Amazon Web Service (AWS)?
Is it possible to configure the same on AWS?

[hsujdik]

I never did that, but you probably would need  to call the AWS API/CLI to assign the Elastic IP to the EC2 instance of the currently pirmary SIP Server on your script.

See the first example of https://docs.aws.amazon.com/cli/latest/reference/ec2/associate-address.html

Edit:
Or, even better, your script could detach an elastic network interface and attach to the currently primary SIP server
https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/using-eni.html

I haven't tested either of these methods, but teorically could work. Give it a try

[gen_den]

Thank you for the reply  will check

[hsujdik]

Just out of curiosity, did you manage to accomplish what you needed?

[gen_den]

Sorry I could Not achieve the same.
I am trying to find other options on AWS

[gen_den]

Just FYI , i tried on AWS but it not achievable.

[Dionysis]

It should work.

I have some eSBC vm's running in AWS which use a similar method for HA failover and it works fine.

Care to share what you've done and what's actually going wrong?

[gen_den]

Hi ,

I created 2 linux instances and then I created a new network interface (ENI) and associated with one instance.
Using AWS CLI i attached new ENI with one LInux instance so now it has 2 IP's  eth0 and eth1 and after switchover I associated this VIP(eth1) with another instance.
This doenst help because i understand we need this same ENI (VIP) to be attached permanently with both the instances.

As per AWS support:
-----------------
The relationship between an ENI and EC2 instances is 1:1 i.e. each ENI can only be used with one instance at a time. Because of this you cannot attach the same ENI with two different instances. With same ENI these instances could face communication error and thus this configuration is currently not supported. Thus, the only way to achieve your use case would be to utilize the attach-detach method.

could you please share how you achieved it?

[Dionysis]

You don't need the ENI associated with both instances at the same time.  Only with the instance which has the primary SIP Server.

So your HA scripts will need to execute the same command as you did when associating the IP with your other instance. Eg. when sip server a becomes primary, the ha reaction scripts will need to disassociate the IP with sip server b and associate it with sip server a (and vice versa).

Sounds like you've got all the pieces, just need to put it together.

[gen_den]

Thank Dionys for the response.
We finally went with Network Device-Based HA.

Configured the network LB on AWS  monitoring the SIP server for health checks and if the SIP server is not reachable, requests start going to another SIP server. It's under testing as of now.

[Dionysis]

What if the Network LB goes down?  Seems like you're removing one single point of failure and inserting another.