Move UCS blade to a different chassis environment

[nini_8]

Hi we have B200 M4 UCS servers, and we also have 2 sites (Same domain, but different Vlan information) I need to move 1 blade from the chassis in 1 site to the chassis in the other site. We also need to rename it, don't know what the steps are. Also, if we still need to reinstall the ESXi OS again??

Thank you

[MDonchatz · Accepted solution]

Step 1) Migrate VMs, power down the ESXi host. Storage  vMotion anything that's locally stored.

Step 2) Disassociate the existing service-profile from the blade. (this step may wipe/scrub local disks - see note below)

Step 3) Decommission the Blade you are removing. Decommissioning removes all info from the UCSM  for a blade that you don't plan to return to service in the same slot.

Step 4) Install the blade in the new chassis, acknowledge it,  and allow time for the chassis to discover it.

Step 5) Reapply a service profile that will work for the new chassis (VLAN scheme etc.)

Step 6) Reinstall ESXi if necessary .....(or untangle / rearrange ) the vmnics at the Hypervisor level.  

A few thoughts/notes:

When you say "same domain" I assume you mean that both chassis are controlled by the same UCSM/Fabric Interconnects?

Thus all the VLAN info would be the same in the domain, however, different VLANs would be applied to the server based on which service-profile was used.

If ESXi is installed on a compact Flash or local hard drive and a local disk "scrub" policy is allied in the service profile, you will likely have to reinstall ESXi.  To avoid this you can change that policy before service-profile disassociation.  Even if you do this, ESXi might get confused when a new set of virtual hardware is provisioned when you apply a new service profile. Thus you may need to check all vDS or  vSwitches to make sure that the proper interfaces (vmnics) match where you think they should.

When you say "rename it" -  You can rename the host whatever you want at the Hypervisor / OS level and UCSM won't care.

You can also create a clone of whatever service-profile that you'll attach and call that whatever you want as well.

Good luck & HTH, please rate posts that are helpful. 

Thank you.

[flamegri11ed]

Dear Community,

Don't know if I am in the correct subforum and hijacking another question, but I have a similar issue.

We had to remove a CPU (8 core) and related DIMMs for Oracle CPU count licensing reasons.

The Blade was reinserted and came up with 14 errors in the UCS manager. (attachment)

What is the correct procedure to fix the inventory issue on the blade? Does this fall under an add or under a removal or under a modify.Not sure what to do in this case. We need to upgrade firmware from 3.2.2 to 4.1 as we are adding an M5 soon and it needs to be corrected before the upgrade.

A shutdown and disassociate the current service profile is the first step. Removing the blade is not the next step as it does not correct the inventory after adding it and powering it on.

What would be the next step?