topic Re: ClusterXL Forced Failover command in General Topics

ClusterXL Forced Failover command

Machine_Head — Wed, 29 Oct 2025 07:27:40 GMT

Hi Team,

Found recently this command:

Configuring Duration of Forced Failover in High Availability Mode

Description

In the ClusterXL High Availability mode, you can force the current Active cluster member to fail over to the specified cluster member for the specified number of minutes.

After the specified duration ends, or after you manually disable the forced failover, the cluster recovers based on the mode you configured in the cluster object:

Primary Up - The cluster fails over to the cluster member that is configured as primary (if it is not currently Active).
Active Up - The current Active cluster member remains Active.

Syntax

ShellCommand

Gaia Clish	set cluster force_failover member_id <ID> {time <Minutes> on \| off}
Expert mode	cphaconf force_failover member_id <ID> {time <Minutes> on \| off}

Could you elaborate on how does it work and what it was thought for?
- Any specific use cases?
What are the differences with admin down?
Can you force a fail-over if the member you are failing to is down?*

* For example recently, after patching R81.20 92 > 118, patched member went down because non patched member was monitoring 3 vlans on a trunk, as opposite to 2 on the patched member. Going down for Interface mismatch IAC pnote.
So i was in need to force a failover and was wondering if this was the command.

Thanks!

Re: ClusterXL Forced Failover command

the_rock — Wed, 29 Oct 2025 15:04:18 GMT

I believe that clish command is same as clusterXL admin down, as far as I know. Can you force failover of member if you are failing to is down? To me, I cant logically see how that would be possible, if its already down.

Re: ClusterXL Forced Failover command

emmap — Thu, 30 Oct 2025 05:11:39 GMT

I haven't tried this so I don't actually know if it will let you do it to a down member. It may, but if it does it will fail back once your time expires, because that cluster member has a problem.

Re: ClusterXL Forced Failover command

Machine_Head — Thu, 30 Oct 2025 09:45:53 GMT

True, but in some cases you still want to failover to a "down" member.

To continue with the example, that member is only down because of a difference in code and thus clusterXL monitoring different interfaces/vlans on the two versions, not because there is a "real" reason for it to be down.

There might be other cases where you still want to failover to a down member, CoreXL differences come to mind..

In my case what i did to force the failover, (and so i could patch that member) was cphastop, but was just wondering if there was a more "elegant" way of doing it.

Re: ClusterXL Forced Failover command

the_rock — Thu, 30 Oct 2025 11:13:00 GMT

In all my years of dealing with CP, I never had to do something like that when patching the member.