Replace VCS Principle Switch

[Russ]

Hi all,

Pretty new to Brocade Switching.

I’ve been given a project to replace both switches in a 2 switch VCS stack with higher capacity switches. The process to replace the non-principle switch looks pretty straight forward (although there are a couple of ‘got-ya!’). My query is about the principle switch replacement. I’ve done a bit of a search on this but couldn’t really find anything to answer my question. Is the process the same as this KB describes? https://extremeportal.force.com/ExtrArticleDetail?an=000061358 Do I need to fail-over the principle switch the the second node first? Are there any other things I need to be aware of when swapping out the principle?

Thanks all. 

[Truyen_Phan]

Yes, the steps are covered in this KB:

https://extremeportal.force.com/ExtrArticleDetail?an=000078756

[Russ]

Thanks for the heads up. Is there a KB on adding an additional switch into an existing VCS fabric?

[Truyen_Phan]

No, you can only run the replacement procedure for replacing the same model, because the configuration from the failed switch contains the port numbering and platform ID within the backend database. So, the hardware needs to be the same. 

This is also mentioned in the management guide: See page 80

https://documentation.extremenetworks.com/networkos/SW/73x/nos-730-managementguide.pdf

Replacing a node in a VCS cluster If a node in a VCS cluster becomes damaged and can no longer be used, a similar node with identical capabilities can be used in its place.

The new node must use the same RBridge ID of the node that is being replaced. When the new node is detected, it joins the cluster as a previously known node instead of being considered a new node.

If it’s a different model, it will need to be added into the cluster as a new RB, then you have to manually restore the configuration once it joins the VCS.  If you want the new RB (different platform) to have the same RB ID as the failed one, then the old RB information needs to be removed and deleted from the VCS before the new RB joins the VCS. 

[tmasuda]

Is there anywhere that says you can replace it with a different model?

I would just try adding it as an additional node in the cluster.