Unable to Form Link Aggregation between HP FF 5700 Switch and VMWare NIC Teaming

[Michael_W]

Hi

I need some advice on link-aggregation issue that I encounter during my setup with VMWare NIC teaming. On HP 5700 switch, I configured a static link-aggregation, the configuration is as below:

interface brigde-aggregation1

port link-type trunk

port trunk permit vlan 100 200 300

undo port trunk permit vlan 1

interface ten-gigabitethernet 1/0/1

port link-type trunk

port trunk permit vlan 100 200 300

undo port trunk permit vlan 1

undo stp enable

interface ten-gigabitethernet 1/0/2

port link-type trunk

port trunk permit vlan 100 200 300

undo port trunk permit vlan 1

undo stp enable

And, on VMWare side, the load balancing policy is set to Route based on IP hash

Somehow, the configuration does not work. Is there anything I need to do at the physical switch side?

Regards

Michael

[parnassus]

What's about permitting - instead of denying - VLAN 1 on the Static Trunk (and, consequently, for coherence, doing the same on any of its member ports)?

As far as I know Port Trunking steps order is important on Comware based Switches: after creating VLANs, after creating the Trunk and assigning member ports to...you should have set (logical) trunk interface's type to Type Trunk and permitted various VLANs on it (VLANs settings should then propagate down to BAGG member ports).

[sdide]

Hi,.

dunno if you just didn't paste it, but you need to add the physical interfaces to the bridge-aggretation interface

for both your interfaces:

interface range Te1/0/1 Te1/0/2
 port link-aggregation group 1
 quit

otherwise they are just individual interfaces.

Regards

[VoIP-Buddy]

Going forward, you don't configure the individual interfaces.  You configure the link-aggregation which then configures the member interfaces.

Why aren't you using a dynamic link-agg to the VMware host?  It will deal with links that disappear for some reason better than static.

Regards,

David

[parnassus]

Dynamic Port Trunking (LACP) against a VMware host requires explicitly that VMware hypervisor uses a VDS instead of the more common VSS...this could mean that *maybe*, if the OP set a static trunk (no LACP) on his Switch, his VMware host uses a VSS.