Restoring a switch after a swapout

[Steven.langdale_2]

Hello all

I'm looking to update our documentation on MDS9000 switch recovery - specifically replacing the whole switch after an swapout.

The MDS cookbook has a procedure for replacing one but makes no mention of what to do differently if the switch is in a larger fabric (and therefore alot of the information is already know to the SAN)

The Procedure basically says, stick an IP address on it, do a tftp restore, save it and reboot.

Is that best practice in a multi-switch SAN?  Or would a better procedure be to restore the switch specific stuff and rely on normal switch operations to replicate all the other info i.e. device aliases and zone info etc.  So in essence, the same procedure for adding a new switch.

Thoughts?

Thanks

Steven

[Kris Vandecruys]

Hey Steven,

The procedure is intended more for replacing a broken switch.  If you go change the model, or if it's connected into a different fabric, things become trickier :-)

If I understand you correctly, a good approach would be to install the new switch in parallel (topology-wise) to the existing one, selectively restore the config to avoid conflicts and rely on CFS to absorb information from the fabric such as device aliases.

Do you have requirements that conflict with this? Things such as keeping the domain IDs identical to the old one, port availability constraints....

HTH,

Kris

[Steven.langdale_2]

Hello Kris

Thanks for the reply.  Infact I was taking about replacing a broken switch.

So as an example, I have a fabric with 5 switches in it, one of the switched dies and gets replaced.

The Cookbook procedure is to restore EVERYTHING, but is that "best practise" in that alot of the info is already on the SAN and should be replicated back via CFS (as you had stated).

So the question is, do you recove the LOCAL config stuff and rely on CFS or restore everything.  Which would Cisco reccomend?

Thanks

Steven

[Nisar Sayed]

If you have local config saved someplace then you should go ahead and restore it to the new switch. Any feature or service related to SW WWN needs to be reconfigured. For instance, features such as Port Security, FCSP (AAA), InterOp Mode 4 (rare case), Cleanup/update your Fabric Manager database. Thats all I can think of right now. Could you please share the root cause, why your original switch went down? Just curious to know.

Thanks,

Nisar Sayed