Issue with setting up networking for vMotion

[mrnick1234567]

Hey,

I'm setting up 6 hosts running vSphere4.1 with my virtual machines on NFS storage.

I have a distributed switch across all 6 hosts. I setup vmKernel0 to use 2 trunked uplinks on each server for NFS storage access on vlan 301.

I then added a second vmKernel port (vmKernel1) for vMotion. This uses another 2 trunked uplinks. I read vMotion can operate on it's own isolated non routable network. So I set it up on an unused VLAN with each of the 6 vsphere hosts having 192.168.0.x addresses.

However trying to perform vMotions resulted in time out errors saying the machines can't contact each other. Also I was unable to leave the gateway blank or set it to a dummy address. (It wouldn't change from the 10.65.x.x address range that the vmkernel0 gateway is set to)

The only way I have got vMotion to work is by putting the vMotion network on VLAN 301 and setting the gateway up to the same as my storage network. However in doing this I am worried my storage traffic might be going over the vMotion uplinks which I don't want.

Can anyone help with this setup please?

Thanks

Nick

[weinstein5]

Take a look at the physical switch ports and make sure they are configured approriately for the vLAN for vMotion - you can only have a single gateway address for the vmkernel so as long as the vmotion ports are on the same physical segment there will be no need ffor a gateway address on the 192.168.0.x network -

[iw123]

once they're on the same physical network/vlan use vmkping to test connectivity.

[mrnick1234567]

Hey Both,

Thanks for the replies. So I've just moved the vMotion ports  back to their own dedicated VLAN with 192.168.x.x addresses. I can ping the other hosts with vmkping and vMotion  now works.

As you say the gateway wants to be the same as  vmkernel0 so I've left it as is. As the vMotion VLAN is only on the  local switch so it won't be able to send packets to that addresss  anyway, so I guess there is no danger of storage traffic using those  ports.