cancel
Showing results for 
Search instead for 
Did you mean: 

Replace a complete switch stack

David31
New Contributor
Hey guys, it looks like we are having an issue with a stack that has 5 nodes. At some point or another a node will fail so it came to be that all nodes will be replaced in this stack. My question is that since I have a limited time frame of 2 hours, would it be best to do 1 node at a time starting from 5 and work my way up to 1 (Master) or do I set up the new stack and copy config files? More than likely the new switch's will have newer image than what we currently have. I've swapped out nodes before but have never been faced with a situation like this.

Any ideas?
9 REPLIES 9

Dave_Hammers
New Contributor III
Drew actually beat me to it. I was going to suggest a combination of cloning to create a USB. Then USBZTP on the new stack master. Then run clone.py slot all to sync all of the slots on the new stack to the master. Available in EXOS 22.2.

I do have a version of cloning, soon to be out, that will clone an entire stack from USB or across a network and is compatible with 22.2 and later.

FWIW, I've posted the clone.py doc to our EXOS applications github under an experimental directory, until the clone.py app download is available.

https://rawgit.com/extremenetworks/EXOS_Apps/master/experimental/clone/docs/clone.html

David31
New Contributor
@drewc. Awesome, now I know moving forward!

Thanks again guys!

Drew_C
Valued Contributor III
This may not be helpful for you now, but I wanted to mention it for those replacing standalone switches (and stacks in certain scenarios), who find this thread in the future.
There's a "cloning" tool in EXOS now:
https://documentation.extremenetworks.com/release_notes/ExtremeXOS/22.4/EXOS_Release_Notes/22.4/c_cl...
https://documentation.extremenetworks.com/exos_22.4/exos_21_1/getting_started/c_cloning_switches.sht...

davidj_cogliane
Contributor
Sounds about right.

If your not changing the config, you should be able to simply save it as a script pull it off and run the script on the new stack. Watch your password encryption if changing code versions.

sho config | include untagged

Usualy works pretty well for identifying non-standard ports. Not sure about your environment but we find most of our customers have 90% of there ports configured on one VLAN and just a handful scattered between things like APs and cameras. Typicly you can just label that last 10% of ports and not worry about where the other go.

Only other thing to watch is ports that have their speed hard set for older devices.