Wired Intelligent Edge

Hi all,

ArubaOS-CX 10.01.0030 Software Images for both Aruba 8400 and Aruba 8320 have just been published.

References:

• ArubaOS-CX 10.01.0030 for Aruba 8400 Release Notes
• ArubaOS-CX 10.01.0030 for Aruba 8320 Release Notes

Enhancements:

• Added Dynamic Inter-VRF Route Leaking (D-IVRL)
• Added a warning on (Unsupported MGTM Interface as) Source Interface
• Added Split Recovery mode feature through the split recovery command into the config-vsx context (to prevent traffic loss when Split Brain scenario happens as when the VSX ISL goes out-of-sync and VSX ISL Keepalive subsequently also fails).

plus varius Bug Fixes.

Hey Parnassus,

maybe you can help me out with the split recovery mode. I saw it's enabled by default but I'm not sure how it works.

From the release notes:

Split recovery mode has been added. Split-recovery mode prevents traffic loss when the ISL goes out-of-sync and keepalive subsequently fails. When the ISL goes out-of-sync and keepalive is established, the secondary VSX lags will be brought down. If keepalive also fails, this causes a split condition. In this case, if split recovery mode is enabled, the secondary will continue to keep its VSX lags down. Split recovery mode is enabled by default

If I have a pair of 8400s and the primary hangs for some reason, ISL will go out of sync and VSX keepalives will stop being sent.

With split recovery enabled (default), the secondary will just shutdown all it's multichassis LAG interfaces? Is this right or am I missing something?

If that's the case, a simple reboot/upgrade of the primary 8400 will cause a service disruption, wouldn't it?

I'm considering upgrading and I want to fully understand how this works and I couldn't find anything on the documenation.

Best regards

2 points:

-1- Explaining a bit more possible failure scenarios for ISL and keepalive:

See picture below. Split-recovery has an effect only during split brain (when split detection is not possible, network admin would choose the end-state)

-2- during upgrade you do not hit this split context.

A rebooting VSX node has to run through a initial sync process for DB synchronization before VSX LAG become forwarding. So there is no interruption as primary or secondary is back online from reboot. The peer VSX node will continue forwarding until neighbor is ready.

Hi Vincent,

that clarifies it a lot. So, on the case you mention, ISL down and keepalive up, with split-recovery on/off all the VSX LAG ports on the secondary are down, so far so good.

If split-recovery is off and keepalive goes down as well, the ports are kept down, so the ports are only brought up if split-recovery is enabled, right?

Thanks you for your help!

Aarón

Correct.

But again during upgrade of VSX pair, you 'll not face a split scenario.

Understood. Do you know if those slides are available for partners and where could I get a hold of them?

Thanks again

VSX technical presentation (10.1) is available here (partners):

https://afp.arubanetworks.com/afp/index.php/ArubaOS-CX_Deep_Dive

We will update it soon for 10.2 release and associated enhancements.