[c-nsp] HP VM ESX fcoe issues with Nexus 5020

[Ryan West]

On Tue, Oct 25, 2011 at 08:45:25, Nyman, Eric wrote:
> Subject: [c-nsp] HP VM ESX fcoe issues with Nexus 5020
> 
> All,
> 
> I'm having an issue with my ESX servers that are connected to our 
> Nexus 5020's using FCOE to connect to our storage MDS9500's. 
> Basically, if for any reason connectivity (either FCOE or Ethernet) is 
> disrupted to the 5K's, the ESX servers will not recover and will 
> require a reboot to reconnect to the storage. Cisco TAC have been 
> looking into it for some time now but they have not been able to 
> provide any information. Cisco's recommendations were to try the 5K's in either NPV or NPIV mode but we get the same result.
> In another scenario, we also had a Cisco UCS chassis that would not 
> connect to the storage unless a shut/no shut was initiated on the 
> switch port. That seems to be resolved with a driver update but only 
> on the 5K switch that is NOT NPV enabled.
> 
> 
> 
> Anyone ever had any experiences with ESX servers connecting to storage 
> on the 5K's?
> 

I had many issues in the beginning with FCoE coming online as an access port with Emulex cards.  Trunking resolved that issue.  As far as recovery goes, I was having problems with the 5010 rebooting on an earlier 5.0(2) code and corrupting portions of my voice lab.  That's been resolved with 5.0(3)N1(1b), which has been running a little over a month with no incident.  During the reload, the hosts were obviously disconnected from storage, but would reconnect when the fabric came back online.  These are the versions I've tested:

C210 M[12] - QLogic QLE8152 10 Gbps 2 port CAN
5010 w/ N5K-M1008 8x1/2/4G FC Module
Hitachi AMS2100 directly attached to N5k-M1008
CIMC/BIOS version from 1.3 to 1.4(1a) currently
ESXi 4.1 Initial release to 5.0

Haven't tested with NPV or NPIV though.

Thanks,

-ryan