rebooting C7000 blade causes switch to reset

[J.Dan]

Hi All,
Has anybody ever seen a problem like this?

I have a node within a Blade C7000 and when I reboot that particular node (call it mynode02) it causes the network to crash. Reboot of any other node in the blade works fine. The problem is just with mynode02.

The the network switches installed are cisco 3020's.

I am just in the initial stages of troubleshooting so please excuse the sparse details.

Just curious what in the world could possibly cause this?

Thanks,
Dan

[Johan Guldmyr]

Nice one!

Does it crash when the server boots or during the logout/shutdown?

If Linux maybe there's a script doing something at either of them.

Is this blade identical to the other blades? In terms of hardware/drivers/firmware/OS-patches?

Any differences maybe on the port in the switch - like STP enabled?

With network crashing - are you running some kind of network analyzer to see what is happening?

Broken NIC?

[JohnDickJr]

Yes we are currently experiencing the same exact symptom.

I have a brand new blade that is in a reboot loop as we have not set it up yet.

Every time it reboots if affects the entire network, not just the local switch.

We are using passthru modules connected via trunk ports (for ESX) to Cisco switches.

We have spanned the port on the Cisco switch and captured a trace (just this past Friday).

Looking at the trace over the weekend and I'm not smart enough to tell what is going on.

We have several good network engineers that will hopefully figure it out.

I will keep you posted,

John

[JohnDickJr]

SOLVED - 

The ports connected to the blades are configured as trunk ports for ESX hosts. When the cisco switch saw a change in link state of the switch/trunk port(s), it initiated a spanning tree reconverge.

We want to keep the ports as trunk ports so we added "spanning tree portfast trunk" to the trunk ports connected to the blades and the issue is resolved.

We Win!!