How to replace a failed switch in a stack?


I have two x460-48t switches configured in a stack. I want to create a procedure for replacing one of them if it fails. I searched and found a similar question from a year ago. But the document that was offered as a solution is no longer available to download. Can someone please help me out?
We will have a spare running the same XOS on the same partition as the active/backup nodes. So we can remove the failed node and install the spare. The necessary commands to add the new node and synchronize the stack is what I'm not clear on. Thanks a lot

13 replies

Userlevel 7
Hello,

Take a look at this article to see if it helps...

http://gtacknowledge.extremenetworks.com/articles/How_To/How-to-replace-a-node-in-a-SummitStack
Hi Doug

Yes that is great! Thanks

Justin
Userlevel 7
No problem, have a great weekend.
Userlevel 1
The following could make life easier when you have to change the master node:
Log into the master-node "run failover" if you have still the possibility. Now the master node becomes the backup node! And proceed like described under the link.
Maybe an idea to think about is to configure stacks always with "configure stacking redundancy maximal" and to give priorities.
conf stacking redundancy maximal
conf stacking slot 1 prio 100
conf stacking slot 2 prio 90
conf stacking slot 3 prio 70
conf stacking slot 4 prio 60
Particularly when they are from the same switch typ.
Then you don't have to think a about the master-capability!
Thanks Jackmikel
Userlevel 2
I would like to add a question to this procedure. I have a 2 switch stack and slot 2 has failed. If I just received a replacement switch and take it out of the box, I know that I need to match the XOS and partition, but after that, do I need to enable stacking on it before I try to add it as a replacement? And what would be the procedure for that part?

The procedure linked above makes it seem like you can simply plug a replacement into a stack and enable it from the master node. Am I missing anything here?

Thanks,
Stan
Userlevel 7
I would like to add a question to this procedure. I have a 2 switch stack and slot 2 has failed. If I just received a replacement switch and take it out of the box, I know that I need to match the XOS and partition, but after that, do I need to enable stacking on it before I try to add it as a replacement? And what would be the procedure for that part?

The procedure linked above makes it seem like you can simply plug a replacement into a stack and enable it from the master node. Am I missing anything here?

Thanks,
Stan
Hi Stanley,

That's exactly right. As long as the EXOS image, partition, license, and stacking protocol match, you can plug it into the existing switch and do the actual stacking config from there.

The only additional change that may need to be done first is enabling stacking-support on any hardware that can be used for either stacking or user traffic (e.g., VIM4-40G4q on the X670V). This would need to be done first, before the switch can communicate with the rest of the stack.

-Brandon
Userlevel 2
Hi Stanley,

I'm facing the same problem like you, but in this case my node that has shown failure is node 1(old master), and after receiving the new replacement node, I'm wondering the same question, if I needed enable stacking support before I plug this one on the stack again.
Userlevel 7
Hi,

you need to enable stacking first.

Erik
Userlevel 2
Hi Erik,

It just If I was using that sort of port "chameleon", or be, where i can use either stacking or traffic, is it rigth?

In that situation I'm using the module SS-V80-B-2, it is exclusive for stack.
Userlevel 7
Hi,

the switch needs to be configured to use stacking before it can be added to a stack. See e.g. How to create a stack with Summit switches.
3. For each switch on which the stacking ports are not already enabled, issue the command enable stacking-support.
After that the correct stacking ports might need to be configured, as well as the correct stacking protocol. Those are prerequisites for the replacement switch to be seen in the stack topology (show stacking).

Erik
Userlevel 2
Hi Everyone,

In order to let all guys informed, We've done the replacement following exactly what is written in that GTAC[¹], and in our case we didn't need to run "enable stacking-support" once we have a specific module for that.
We just changed the protocols from standard to enhanced, and everything worked fine as expected.

[1]https://gtacknowledge.extremenetworks.com/articles/How_To/How-to-replace-a-node-in-a-SummitStack

I hope that info can help other guys as well.

Cheers,
Welisson
Userlevel 1
Hi All

Is it possible to update the FW on the replacement slot using the fw present on the stack?

Thanks

Reply