10-12-2016 05:26 AM - edited 03-01-2019 12:55 PM
Hi All,
We are having a plan to do an upgrade UCS infra and UCS blade servers with Hyper V hosts. We are upgrading from 2.2(5b) to 3.1(1k). Would like to know should the Hyper-V host have latest patches before we do the firmware upgrade on these hosts. Is there any recommendation on this. Or it does not matter if these hosts have or not have the latest patches.
10-12-2016 08:20 AM
Hi
The step from 2.2.5b to 3.1.1k is reasonably small; therefore I would do first
Upgrade the infrastructure (UCSM, IOM and FI)
Then test one blade (having a 3.1.1k loaded Host firmware package) and update OS with the enic/fnic drivers according to interop matrix:
http://www.cisco.com/web/techdoc/ucs/interoperability/matrix/matrix.html
I would think that you won't have big trouble, and then could upgrade all your remaining servers.
PS.
One could also open a preactive TAC case, asking for a impact analysis.
10-12-2016 08:58 AM
Thank you Walter for the response.
I normally follow the procedure of upgrading UCSM first and then the Secondary Fabric Interconnect which automatically updates the IOM and then goahead with the upgraade of Primary Fabric Interconnect which was successful before.
Is it a best practice to upgrade UCSM and then IOM and then Fabric Interconnects. Please clarify
10-12-2016 09:20 AM
Hello,
Here is the guide for manual upgrade:
http://www.cisco.com/c/en/us/td/docs/unified_computing/ucs/sw/upgrading/from2-1/to2-2/b_UpgradingCiscoUCSFrom21To22.html#concept_964464BA5D4847B5897729A64F39F40C
In summary:
1. You upgrade UCSM
2. You Upgrade and activate your IOM, ensuring to select "Set Startup Version Only". This will put the IOM in a "Pending Next Boot" Status.
3. You activating the subordinate FI, reboot it, and after it comes back online, it tells the IOM that is attached "Okay, now you reboot and upgrade"
4. Cluster lead change after all faults have cleared for subordinate FI.
5. Conduct the same procedure on the other side.
HTH,
Wes
10-18-2016 07:56 AM
Thank you Walter and Wesley for your inputs.
10-26-2016 06:45 AM
Hi All,
Unfortunaltely the upgrade did not go through when we upgraded the sub-ordinate fabric interconnect after upgrading the UCSM we lost access to all the Microsoft HyperV servers which were configured for iSCSI boot. We did check that there are 2 paths and MPIO is configued in the windows server not sure what can be the root cause of the issue. We had to rollback the upgrade and do a hard reset of the server to get the servers back from the BIOS Mode. Let me know if anyone of you encountered this before
Also I do have a question for iSCSI intiatiors names on the 2 vnic though same has a suffix of.a and .b. For ESXi I remember using the same iSCSI initiator name for both the vnics using the IQN pool can this be a issue
10-12-2016 09:45 AM
In manual upgrade (I would do autoinstall), UCSM is always first, because it's the only component which guarantees backwards compatibility !
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide