UCS fabric-interonnect setup issue

niranjandas · ‎11-17-2011

Hi,

I am kind new to the UCS configuration arena. I got a complete setup with a Blade Chassis and two N10-6100 fabric interconnect to it. The setup was used earlier. When i brought it to our lab i need to change the ipaddress that it previously has. I connected through the console, but the cli seems very different from Nexus 5000 and MDS 9500 series switches from cisco. I could not even change the ip address. If anyone kind enough to show me some the way or some documentation i will be greatful. Fundamentally i want to change the ip address first and then some thing more.

I am giving some info from the setup that i have.

cisco-A-A(local-mgmt)# show version

System version: 1.4(1i)

I collected a UCSM CLI guide from net, where it says to change ip address use the following steps:

scope fabric-interconnect a

set out-of-band ip ip-address netmask netmask gw gateway-ip-address

When i tried on my system, it gave the following error.

cisco-A-A# scope fabric-interconnect a

Software Error: Exception during execution: [Error: Timed out communicating with DME]

Additionally i looked at the cluster state too.

cisco-A-A# show cluster state

Cluster Id: 0x77b2297e0a0711df-0xbfe7000decf4ca84

A: UP, ELECTION IN PROGRESS (Management services: UP)

B: UNRESPONSIVE, INAPPLICABLE, (Management services: UNRESPONSIVE)

INTERNAL NETWORK INTERFACES:

eth1, DOWN

eth2, DOWN

HA NOT READY

Peer Fabric Interconnect is down

No device connected to this Fabric Interconnect

WARNING: Failover cannot start, device configuration is incomplete

These are some info from my switch. If some one throw some light, it will be very helpful.

Thanks,

Robert Burns · ‎11-17-2011

In order to change the IP of an existing HA UCS system, the DME & HA cluster must be online. For this you need the cluster links (L1 & L2) connected between the two 6100's and a Chassis connected with at least one link to each IOM.

One this is all connected, your cluster state should show you "HA Ready" and you can then change your IP as detailed above.

Regards,

Robert

niranjandas · ‎11-17-2011

Thanks for the info on the topic. Could you pleas add the following info if you are aware ?

1. If i change only ip address am i good?

2. There seems to be 3 ipaddresses for mgmt port, 6100-A, 6100-B and virtual mgmt ip. Do i have to change all 3 ? Also which ip ( i mean 6100-A mgmt, 6100-2 mgmt or virtual ip) is used to communicate to the UCS system?

Thanks again.

Robert Burns · ‎11-17-2011

IF you want to have an HA system (dual FI's) then you need 3 IPs - Fab Interconnect A, B and the Virtual IP. Assuming your cluster is online, you can change all 3 addresses to the approrpriate subnet and you'll be fine. You'll likely need to do this via the console as your may not able to reach the VIP via IP in the new subnet yet.

For HA mode, each FI has a management IP, and the VIP is used for UCSM access which floats between either FI A or B depending on which one is primary.

For standalone mode, you only need one address.

Robert

niranjandas · ‎11-17-2011

Again thanks for the info provided. What is the exact command for changing ip address? Does UCSM stands for UCS manager?

Robert Burns · ‎11-17-2011

Yes. You had the commands correct above. You scope to each FI and set the out-of-band management IP.

You can set the cluster IP (VIP) by scoping to "system" and setting the virtual-ip.

There's a couple posts on here detailing it, search my name and you'll find it.

Robert