swapping out 2104XP IOM's

[russ.givens]

I got some thermal alerts from a 2104XP IOM last week.  I investigated and saw no thermal issues in our datacenter, nothing was blocking airflow, and not other components were experiencing any type of thermal issue.  So I opened a case with Cisco TAC.  The looked through some tech support logs I sent them and indicated that the IOM's need to be replaced.  I asked if this was a disruptive process and they indicated - in theory it is not.  That answer didn't make me feel all that confident in replacing them.  Has anybody replaced both IO modules in a chassis before.  It is my understanding to just unplug the cables and pull out the IOM, insert the IOM, monitor it's progress in discovery and firmware update / downgrade, and once it's backup to proceed with the next one.  Is there anyway of validating that an IOM is appropriately functioning?  Or are there any specific steps I'm missing.

It's impossible for me to take an outage to do this change.  The chassis hosts several hundred production virtual machines.             

[Gregory Scarlett]

Hi Russ,

Provided your blades have NICs and HBA's connected to both Fabric's, there should not be any disruption.

Replace the IOM's one at a time.  Make sure you wait for the IOM to come fully online - it may take a while as it will probably need to upgrade/downgrade it's code to match the FI. 

Don't move on to the second IOM until the first replacement comes up as Operable, and check the code version it's running to make sure it matches the Fabric Interconnect.

This process can take some time - be patient.  Generally it's about 15mins, but I've seen as long as 30mins.

[russ.givens]

thanks, I will give it a shot.

[Gregory Scarlett]

Russ,  The TAC case is still open, so I'd suggest calling us just before you make the change.  We can have an engineer check over the system before you make the change to ensure that HA to the blades is configured and then assist if there are any problems.

[Keny Perez]

Russ,

Is definitely better if you call us before the change so we verify that your environment is fully redundant.  Remember you may have, or not, configured " Failover " at the service profile level for ethernet and " multipath " for FC, so we definitely better take a look before telling you "it is safe"

My 2 cents...

-Kenny