Trunking Clarification... and SW - Server connections

Unanswered Question
Jan 19th, 2009

Greetings,

I'm trying to understand how a 4507 handles load balancing OR if it's even load balancing when i have 2 trunk links going to a server.

Both Interfaces are trunking and in STP both shows as FWD. So I'm a bit confused because I'm expecting one of the links to be in blocking. Why is this the case? Am applying the wrong theory to this problem?

Any thoughts or recommendations on how to connect multiple trunk ports to server and load balance without using Etherchannel.

STP - pvst

show spanning-tree blockedports - 0

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 4 (1 ratings)
Loading.
Edison Ortiz Mon, 01/19/2009 - 12:14

You will see one port being blocked while the other port forwarding once the switch receives the same BPDU via different interfaces.

As the server does not send BPDU, both ports will be in FWD mode.

With that said, the server has 2 NICs connected to the switch, thus from the switch's perspective is treating these connections as 2 independent connections.

Traffic originated from one NIC will return to that NIC, same theory applies to the other NIC.

If you wanted bandwidth augmentation among both NICs, then you will consider etherchanneling on the switch and teaming the NICs at the server side.

HTH,

__

Edison.

merryllem Mon, 01/19/2009 - 13:41

Edison,

Thanks for the input. The problem I have with Etherchannel/NIC teaming is that when the server looses connection the SW will put the interfaces in a "suspended" state and the only way for the connection to reestablish is for me to shut/no shut the Port-Channel interface. Is there a way around this?

Edison Ortiz Mon, 01/19/2009 - 13:58

It seems they are going into suspended state due to some kind of err-disabled.

If you check the logs in the switch, what's the reason of the err-disabled detection?

Ideally, you should configure LACP in the switch and server instead of relying on unconditional etherchanneling on both link partners.

HTH,

__

Edison.

merryllem Mon, 01/19/2009 - 15:08

Below is the error log:

PST: %EC-5-L3DONTBNDL2: Gi7/38 suspended: LACP currently not enabled on the remote port

I have both sides SW/Server in LACP Active. Which is the recommended configuration given by Cisco. I'm guessing the interface is suspended when the server is in the early stages of the boot up process wherein the NICs are up but not responding to LACP from the SW.

Edison Ortiz Mon, 01/19/2009 - 15:27

Can you try setting the switch as passive and the server left as active?

The switch won't try to initiate the etherchannel until the server is ready.

__

Edison.

Edison Ortiz Mon, 01/19/2009 - 18:02

Do the etherchannel on the switch ever acknowledge LACP at the server side?

Is this a timing issue?

It could be the server isn't configured for LACP properly after all or the switch isn't receiving LACP packets from the server.

Try going with unconditional etherchannel (mode on) on the switch and server and see if it helps.

__

Edison.

Actions

This Discussion