Load balancing settings, and loss of communication

[MagicRecon]

Hi,

I have a Vmware cluster with 6 ESX 4.1 blades distributed over 2 IBM bladecenters.

When all 6 settings for load balancing are on the same setting, eg IP hash, i have connection between all 6 blades. When i change the setting on 1 or 2 blades, the changed blades can interconnect but lose connection to the other blades.

I have 2 physical cables between both blade centers wich are both on the same vSwitch (but on different pSwitches). (VmNic0, VmNic1). When i use explicit failover i can use both nic0 and nic 1 as primary nic, but i cant mix nic0 and 1. So i think there is a problem making connections wich go out on one NIC, and comming back on the other NIC in the team.

From a network point of view i should be able to make connections going out from nic0 and comming back over nic1, because they meet on the same vSwitch. but this is not the case.

[admin]

ip hash teaming should work in conjunction with etherchannel on physical swithes.

[AndreTheGiant]

Welcome to the community.

I suggest to start from the simple policy (Port ID) to make your test.

In this configuration you need only a good STP configuration and nothing more.

As written with IP hash you must configure in the right way also the switch ports.

Andre