Chassis/FEX Discovery Policy Global Policy Change

[Sgiesbrecht]

Hello All,

I have 2 6332 -16UP Fabric Interconnects with 3 chassis connected. I am trying to connect a new C240 M4 to my fabric interconnect via a 40gb port. The port has a fault because the chassis/FEX Discovery Global Policy is set to 4 Link. The new C240 I am trying to add only has 1 link. So my question is if I change the policy to 1-Link Discovery Policy will this cause a disruption? AS well my current chassis have 4 connections will this cause disruption or loss of service?

Thanks,

Scott

[Kirk J]

That policy does not impact rack server discovery.

I would not adjust that in troubleshooting your C series integration troubleshooting.

I would make sure all the firmware for rack server including adapter meet minimum requirements, the adapter is in the correct PCI-E slot that allows single wire mgmt, etc.

Thanks,

Kirk...

[Sgiesbrecht]

The fabric interconnect currently gives the following error for the port I have connected the new c240 to:

F0277

ether port 1 on fabric interconnect B oper state: link-up, reason: FEX not configured

The Cisco UCS Manager discovered the chassis with only one link in the first attempt although the chassis discovery policy is configured for four links. Until the discovery is successful with that link, the remaining links cannot be marked as active.

This is why I ended up at the global policy setting. I had initially connected the c240 to a 10GB port on the FI and it was recognized and brought in to the UCS manager. Now we are changing the port to a 40GB port and it is not connecting.

Thanks,

Scott

[Wes Austin]

Can you confirm the cable optics you are using to connect the C240-M4 to the 40G port?

Are you using VIC 1227? It sounds like you might be using VIC 1385?

[Sgiesbrecht]

I am using the 1227 with the adapter to go from 10GB on the the FI to the 40GB connections.

[Kirk J]

Did you decom the rack server, prior to switching out the cabling from 10 gb to the 40Gb adapter?

You may need to go into CIMC/F8 option on console/crash cart, and reset (reset to factory defaults) the CIMC, so discovery can be completely started fresh via the different 40Gb VIC card.

thanks,

Kirk...

[Sgiesbrecht]

I did not decommission the server before changing the connections. I will try that and I will reset the CIMC.

Thanks,

Scott

[Sgiesbrecht]

I have decommissioned the server and reset the CIMC to factory; however, I am still getting an error on the port in the FI. The error I am getting is:

Description:ether port 1/20 on fabric interconnect A oper state: failed, reason: Unknown

ID:3776934

Type:network

Cause:port-failed

Created at:2018-06-14T09:21:53Z

Code:F0277

Number of Occurrences:2

Original severity:Major

[Wes Austin]

Have you tried to change the speed from 40G to 10G on the port?

[Sgiesbrecht]

No I have not but, I wanted to take advantage of the 40GB speed. I'm planning on pushing a few hba's and nic's through this connection. I will try it and see what happens.

[Wes Austin]

your previous post "I am using the 1227 with the adapter to go from 10GB on the the FI to the 40GB connections"

The 1227 is a 10GB, so it would need to negotiate with a 10GB link on the 6300. This would require a QSA and the port speed to be 10G on the FI.

[Sgiesbrecht]

To change the speed it would need to be configured as a breakout port correct? I have a QSA connected to the 40GB port.

[Wes Austin]

Can you send the following outputs from the problem interface:

show run int x/y

show int x/y

show int x/y trans detail

[Sgiesbrecht]

I am trying to run the commands you have asked about; however, I'm new to the cisco CLI, How do I get into the right context to show the int?

[Wes Austin]

When you SSH, just "connect nxos" and then run them

UCS#connect nxos

UCS(nxos)#show int ....