LACP blocked ono switch after server reboot

[BOVC]

Hi,

I've got a couple of VMware ESX servers connected to a HP 4506zl switch. It has all been set up for LACP, and this works fine.

But... If I boot one of the servers, the ports on the switch stays in LACP Blocked mode, even after the server is running again.

LACP is configured as trunks, but I've tried to configure the individual ports as passive, active etc. before configuring the trunk, but it doesn't help.

Anyone know of this issue? And by the way, the NICs on the IBM servers are broadcom and Intel.

Thanks in advance,
Rasmus

[Matt Hobbs]

I've heard of this one before... I believe what is happening that on the reboot, one of the NICs goes to 100FDX which breaks the LACP trunk (different speeds).

To fix this, set the server ports to 'speed-duplex auto-1000'

Don't forget to assign points to any replies you receive.

[BOVC]

I've now set both the servers and the switch to static gigabit speed. This doesn't help however :(

[Matt Hobbs]

Weird, instead of using LACP, maybe try setting just the standard trunk type on the switch:

e.g. trunk a1-a2 trk1 trunk

They both use the same load balancing algorithm so I believe it should work (I expect it would work between 2 switches but not too sure if a server will like it, still it's worthwhile trying).

[BOVC]

I believe that will work. Only thing is.. I'm not interested in using regular trunks.

The server are running VMware, and LACP is much better ditributing traffik across the links, than the "protocol-less" trunk links.

So I'm stuck :(

[Matt Hobbs]

Do you have any different switch models you can test with (5300 for example) to see if this specific to the 5400 or not?

LACP and Trunk on the switches use the same load balancing method, so if you can leave LACP on the server and set Trunk on the switch - there should be no difference (as long as the server doesn't complain).

[Geoffrey Day]

I have the same issue.

We have ESX Servers and each has 4 teamed ports using LACP.

We are running the following switch:
ProCurve J8698A Switch 5412zl
Software revision K.12.43

The trunk for each team of ports has been configured on the switch like this "trunk a1-a4 trk1 lacp"

a number of vlans have been tagged over the trunk, and one vlan has been untagged.

When an ESX server reboots and comes back up the interfaces are up but no traffic flows through them. if I do a show vlan command the vlan info says it is down.


Port Information Mode Unknown VLAN Status
---------------- -------- ------------ ----------
Trk1 Tagged Learn Down


To get it working again I remove all the interfaces from the trunk and add them again, followed by configuring the vlans again.

Any suggestions?

[ande]

Hi,
VMware ESX doesnâ t support dynamic LACP, you need to set the trunk type to Trunk.

Regards

Ande