topic Re: HA Active/Active - Failover issues in General Topics

HA Active/Active - Failover issues

darkfibre — Sat, 05 Nov 2011 09:26:21 GMT

Hello,

I have tried to setup an Active/Active cluster, based on the "PAN-OS Active/Active High Availability - Configuring active/active clusters" technote. I am pretty sure that I have understood and followed the guide correctly.

However, regardless of the Session Owner, Session Setup and Virtual Address settings, the cluster did not seem to failover correclty.

In each test I have disconnected an ethernet link from the Primary PaloAlto device. The results varied, depending on the HA settings:

a) the failover did not happen at all

b) the virtual address 'failed-over' to the Secondary device correctly; However, after ~20 seconds the Secondary device stopped forwarding traffic from this segment. It resumed forwarding traffic after another 10-15 seconds. Then it stopped again - it kept repeating.
All the while, no additional changes were made to the physical connections.

c) the virtual address 'failed-over' to the Secondary device correctly, but only after half-a-minute or so.

I have also done a test, where I suspended the Primary device. The traffic 'failed-over' correctly, but PaloAlto device seemed to randomly stop forwarding it for short periods of time (situation simmilar to point 'b' above).

During the tests I have been observing the MAC address tables on switches and ARP tables on hosts - it seemes that the Gratuitous ARP packets were sent out correctly by the PaloAlto devices.

While using the same topology and devices with Active/Passive configuration, I had no problems whatsoever and failovers happened correctly with no issues.

The tests were carried out using two PA5050 devices, using 4.0.7 and 4.1.0 software versions.

Has anyone successfully deployed and tested Active/Active cluster?

If so, could you please share the High-Availability part of configuration?

Kind regards,

Re: HA Active/Active - Failover issues

bpappas — Wed, 09 Nov 2011 08:16:28 GMT

@darkfibre:

We have numerous customers with successful Active/Active deployments. Without looking at the specific settings of your test scenarios I would be reticent to offer up any reasons behind the failures that you describe.

If you would be willing to share the configuration settings of your test environment(s) in this thread we can probably figure out if you have encountered a bug or a configuration mishap.

Or if you prefer you can open a case with the support team to get to the bottom of the problems that you describe.

Thank you,

Benjamin

Re: HA Active/Active - Failover issues

commcord — Thu, 10 Nov 2011 18:02:19 GMT

Hello darkfibre,

Have you been able to find a solution to the issues with Active/Active configuration?

Have a good day.

Re: HA Active/Active - Failover issues

darkfibre — Thu, 10 Nov 2011 18:10:35 GMT

@bpappas, @commcord

Hello,

Still no luck, unfortunately.

I have attached a file with the lab topology that I used for testing, as well as the HA configurations for both devices:
- Active/Passive,
- Active/Active with one floating IP per segment,
- Active/Active with ARP load sharing.

Using the same topology for all tests, Active/Passive setup worked flawlessly and failovers happened as expected.

However, Active/Active configuration didn't seem to work correctly for me, regardless of the Session Setup and Virtual Address settings.

Please let me know if anyone finds an issue with the above configurations.

Kind regards,

darkfibre

Re: HA Active/Active - Failover issues

jnguyen — Sat, 12 Nov 2011 00:36:34 GMT

We reviewed your configuration and it looks ok. What we would like for for a debugging session if possible. what i recommend is to call into support and open a case with support to debug this issue further.

Re: HA Active/Active - Failover issues

darkfibre — Sun, 13 Nov 2011 10:13:02 GMT

@jnguyen

Thank you for your answer.

I have opened a support case just as you suggested.

Will let you know if the problem is resolved.