VSX Failover Logic

[Malik1]

Hi Experts, We plan to deploy two 26000- gateways in VSX . I wanted to understand the failover logic when Chassis HA mode. In case a VS fails , lets say that the interface that is assigned to the VS that connects to as distribution switch fails . Does the failover only happen for that VS or ALL VSes will fail over, not just one. If case the answer is YES . can we have individual VS failover in HA mode . Dont want to use VSLS. Regards, Sijeel

[Chris_Atkinson]

R81.10 JHF T75 fixed a related issue with Bond interface failure.

HA is actually facilitated with VSLS now days.

Note:

The only mode available for a VSX Cluster that was installed as R81.10 and higher is Virtual System Load Sharing (VSLS).

The High Availability mode is available only if you upgrade a VSX Cluster from R81 or lower to R81.10.

[Malik1]

Chris,

So if i do a fresh installation with r81.10 or high the only option i have is VSLS.  Do u have any documentation or SK that i can refer to. 

Thanks 

[Chris_Atkinson]

The source for the above statements is the R81.10 VSX admin guide, please refer:

https://sc1.checkpoint.com/documents/R81.10/WebAdminGuides/EN/CP_R81.10_VSX_AdminGuide/Topics-VSXG/V... 

[_Val_ · Accepted solution]

Would R81.10 VSX Administration Guide be enough?

Quoting from there:

Virtual System Load Sharing (VSLS):

• Ensures continuous operation by means of transparent VSX Cluster Member failover.
• Enhances system performance by distributing Active Virtual Systems amongst VSX Cluster Members.
• Efficiently balances network traffic load by distributing active Virtual Systems between VSX Cluster Members.
• All VSX Cluster Members and Virtual Systems are continuously synchronized.

Note - This is the only mode available for a VSX Cluster that was installed as R81.10 or higher.

[_Val_]

@Chris_Atkinson Just to clarify, if a cluster was in HA mode and then upgraded, it will still be able to run HA, also in the guide. I just do not see the point of keeping it that way 🙂

[Chris_Atkinson]

@_Val_ Don't worry I 100% agree, HA is now a subset config of VSLS exactly as you describe. 🙂

[_Val_ · Accepted solution]

First, let me say, VSLS is the preferred mode for everything, regardless. 

Classic HA mode means that you run all VSs active on one member, and they all fail over if there is something wrong, event with only one of them.

If you want a failover per VS, this is VSLS. And with manual distribution, you can have all VSs active on your preferred cluster member. VSLS allows that too while keeping per VS failover if something goes wrong. 

[Malik1]

Thanks Val , that sound like a good idea to setup the VSX  in VSLS mode and manually make all the VS's active on one cluster member. And that way i can have failover per VS. 

[_Val_]

Correct.