HP FlexFab 20Gb 2port 650FLB Adapter Crashes Hyper-V Host when Live Migrating VM with vFC Adapte

[mikldk]

Hello People.

We got a enviroment consisting of the following :

HP C7000 Enclosures equipped with 2 pair of HP VC FlexFabric 20/40 F8 Module running Firmware 4.45

HP BL460c G9 with 2x E5-2650 v3 @ 2.30GHz,  512GB of memory and a HP FlexFabric 20Gb 2-port 650FLB Adapter.

OS : Windows Server 2012 R2 w. Hyper-V Fully Patched OS, and HP SPP 2016.10.

So, our problem is that when assigning a VM with virtual fibre channel adapters, and enabling NPIV on the 650FLB Adapter it crashes the destination host when moving this one VM from one Hyper-V host in the Cluster, to another.

We run Microsoft MPIO and IBM SDDDSM 2.4.7.0 software towards our IBM SVC, and when running a enviroment without vFC adapters on VM's in the Cluster, the Cluster is rock solid.

So, anyone else experiencing this?

[GIVI-IT]

We are having same issue running VC FLex 20/40 F8 with 650FLB adapters on 460blc blades.

They keeps saying upgrade firmware but nothing is resolving issue. Let me know if you find the fix for it.

[peyrache]

Shall give me
Os version
Vc version
Firmware/driver version for 650
Thanks
JY

[GIVI-IT]

650flb on host firmware 11.4.1231

Driver ver for HBA FCOE is 11.4.1220

We tried downgrading upgrading nothing works even upgrading SPP for Interconnects and enclousure.

As soon as virtual ports are turned on it crashes the driver on host it loses targets for host and seems like vports take over physical ports. From VC prospective it shows vport WWPN on switch but target storage cannot be recovered on host until it's restarted.

Host OS: Windows server 2016 Datacenter

[peyrache]

What are VC firmware version ?

[GIVI-IT]

Right now its SPP 2018.03 VC firmware 4.50

[peyrache]

Shall you try to update to 4.62 please
There are some enhancements for path failovers
Thanks
JY

[jagilbertvt]

I'm having similar issues w/ 650FLB on Gen10 blades running Server 2016. I've applied SPP2018.03, which includes the latest drivers (11.4.1205.0 for ethernet and 11.4.1220.0 for fcoe).

OneCommand is very slow to discover, and in many cases I see no LUNs connected unless I issue an FCOE reset. This may be on one or both ports.

If I try to live migrate a guest w/ vFC adapters from one host to another, it will fail, though I will see tho WWPN's login to our storage array. In some instances when live migrating the guest will fail completely causing an outage..

I have just upgraded to VC firmware 4.62  and the issue persists.

[mikldk]

Maybe this is solved when running Server 2019 with newest HP SPP - have you tried that?