Comware

Hi there,

I have an issue whereby testing IRF stacking config; I powered down master and the second switch then became the master (correctly so) then when switch 1 (original Master) came back up it did not comeback as the master and switch 2 became the master. 
I never seen this before as I then had to reboot switch 2 for switch 1 to go back as Master.

All switches have the correct priority and IRF config is correct just wondering if I am missing something?

Incidentally, here is the how I have IRF timer etc set;

irf mac-address persistent timer
 irf auto-update enable
 undo irf link-delay

I look forward to hearing back from you! 

Hi @prodigy811 !

Yes, the only thing you are missing is the IRF master election process step 1 - current master stays master. In other words this process is not preemptive. When the old master returns to the stack its priority means nothing, because the stack already has master. In case if current master with lower priority reboots, then the switch with highest priority will become master. It was always like this from the very beginning or IRF protocol. The reason is stack stability. Just imagine what will happen otherwise if a IRF member with highest priority will enter an endless rebooting cycle. Then the whole stack will become unstable as the master will change every 10 mins or so. I hope now you see why this design decision has been made.

Hi there,

Many thanks for getting back to me so quickly. That makes sense, so essentially the priority means nothing except only with initial setup? So, for arguments sake i will have to reboot the new master each time to revert it back to the original master?

Kind regards

I think this excerpt from the guide will be useful:

That is very true - the only way to force another elections is to make current master disappear from the stack (first arrow on the screenshot) Reboot is the way. Some old chassis-based switches running Comware 5 had master failover command, but Comware 7 does not offer it anymore.

And the second arrow proves my statement why your old master does not become new master after re-joining the stack.

Hi there,

Many thanks for your reply it makes a lot of sense. I remember the master failover command, I remember that with previous versions.

Thanks again