Disruptive upgrade to UCS 4.1(2a)

[fvilavella]

We proceed to update from version 2.2 (8i) to 4.1 (2a) through automatic installation with the infrastructure firmware option.

The entire procedure is performed as expected and once the infrastructure upgrade (UCSM, Fabrics, IOM) is finished.

But later we see that each Blade is in discovery mode even though they have User Ack policy. 

Subsequently, we see in the events of each server that apparently the discovery restarted the blade but at the Vmware level (vmware 6.5 cluster) the uptime of the hosts that did not reboo. However, all virtual machines were restarted due to communication failures with the storage (this communication is redundant using both fabric interconnect)

This platform had already been updated to version 2.2 (8i) in Jan-2018 in an equivalent way without problems.

Has any of you something similar happened to him?

[Priyanka L]

Hi there,

Can you confirm what is the firmware running on the blades in the domain ?

If you are using 62xx FI, the minimum firmware on each of the blades got to be 2.2(8) to be cross compatible with 4.1(2) on the FIs.

If you are using 63xx FI, the minimum firmware on each of the blades got to be 3.1(3) to be cross compatible with 4.1(2) on the FIs.

Any firmware below that on the blades, can lead to a communication gap between FIs and blades/ and unexpected reboots.

Ref : https://www.cisco.com/c/en/us/td/docs/unified_computing/ucs/release/notes/CiscoUCSManager-RN-4-1.html#concept_u3g_211_qjb

Search for the keyword "Cross-Version Firmware Support"

[fvilavella]

Thanks for you response, yes we validated before the upgrade "cross-version firmware"

We have firmware 2.2(8a) in the Blades and Fabric Interconnect model 6248UP