VSX Cluster. High Availability

[Vanesa_Benito_O]

Hi all,

I am currently evaluating a VSX environment where no Virtual Systems have been defined, and all traffic and network configuration is implemented directly on the VSX cluster itself (vsenv 0).

Although this is an unusual design (it would likely have been more appropriate to deploy a simple ClusterXL), the cluster is operating correctly. However, I have a question:

In a traditional ClusterXL deployment, there is an option in SmartConsole to define which member should remain active after a failover and subsequent recovery.

This option does not appear to be available in the VSX cluster object.

Is there any way to configure this behavior in a VSX environment?

Best regards.

[JozkoMrkvicka]

In case of VSX VSLS, there is an option to toggle VSLS mode between Active Up and Primary Up using "vsx_util vsls" command. Active Up mode will try to preserve state of currently active member, while depending which member you select as Primary Up, this member will be active all the time (if possible).

[Vincent_Bacher]

In addition to this, you can switch between VSLS mode and classic HA using 

vsx_util convert_cluster

[Vanesa_Benito_O]

Currently, VSX_UTIL it is configured to operate in HA mode, but I am concerned that this setting may only apply to the defined Virtual Systems and not to the VSX cluster itself.

So, when Node 2 goes down, the service fails over to Node 1, and when Node 2 recovers, the service remains active on Node 1. I dont know if I convert the cluster to VSLS I could define the service back to node 2 again, because the V0 doesnt appeared in the configuration...

[_Val_]

Physical cluster members' HA status does not really matter for VSX, since no production traffic is going through VS0

[Vincent_Bacher]

Running VSX without VS is unusual. Is there a reason for this? Have all VSs been decommissioned?

Or was it never planned as VSX?

Depending on the configuration, it might be possible to switch back to classic HA.

I mention this because it could be very hard work to prepare for and carry out the migration, so it might be better to leave things as they are due to the amount of effort involved in the conversion.

[Vanesa_Benito_O]

I’m not sure about the reason, I found the deployment configured that way. I recommended changing it to a traditional cluster, but as you mentioned, it’s part of a project and we prefer to wait until the hardware renewal, which is already planned for the near future. 🙂

In the meantime, we would like to determine whether there is any way to define the preferred active member in that scenario.

[_Val_]

In VSX, the default and, more importantly, recommended mode is VSLS, where an HA pair is defined per VS, and not on the physical cluster level.

You can fine-tune it with vsx_util command, but not on the SmartConsole. It is properly documented; please look it up in the admin guides.