Networking on cluster, am i doing this right?

[mozer__]

Hello,

I have a lab with the following configuration

3 esxi hosts

Vcenter installed on W2008 R2

2 NFS shares on different servers

I've created a DRS/HA cluster and on every host mounted the two NFS shares, also added 3 vswitches with vkernels like this

Switch0 for management (192.168.1.1-2-3 - default switch)

Switch1 for Vmotion (192.168.2.1-2-3)

Switch2 for High availability (192.168.3.1-2-3)

In order to not mix the traffic every swich uses different networks, the problem comes when i try to migrate a VM, i receive the message "can't connect using Vmotion network" and when i connect using ssh the host i can't ping the others in the same network range ej, 192.168.2.2 won't ping 192.168.2.3 and so on)

I even tried to change the vmotion ip's (192.168.2.XX to 192.168.1.XX) so they going to be in the same sunet but i received the same error even when ping respond.

My QS is...the configuration is correct? is valid to use different networks this way? how can i route between networks?

EVC is activated and DRS storage is also set up.

Thanks in advance.

[marcelo_soares]

You need to check why the 2.2 don t ping the 2.3, in the original setup (why all the 192.168.2.x don t ping each other). Can you try to connect a crossover cable between the 192.168.2.2 and 192.168.2.3 to check if they will ping? Also, user "vmkping", and not "ping". Double check if the vMotion interfaces have the "vMotion" checkbox enabled (and only the vmotion interfaces, no other).

The other IPs ping each other?

[mozer__]

Hello Marcelo,

I'm using a lab with Vmware workstation to virtualize ESXI and Vsphere so there's no real switch, all my vkernels have the same gateway (192.168.1.1) maybe the reason the networks don't ping each other is that they i don't have a router that route between them, at first a thought that vcenter does that but now i see that probably i need a routing device... you agree with that?

Now, the config should work if every vkernel (vmotion and fault tolerance) are in the same network (192.168.1..XX) but it does not!!, a get the same error, "vmotion network not available" and if i try to ping the other hosts it respond.

My qs is, does this config make sense?, is this the recommended way to configure a cluster with Ha and vmotion?, maybe setting a VLAN could solve this...

Thanks

[Rubeck]

Are these hosts installed from scratch? Or have you simply installed just one and then cloned it? There might be MAC issues if you did this.. 

/Rubeck

[mozer__]

Hello Rubeck,

The hosts are installed from scratch, but what i really want to know is if what happening to me is normal or not, i mean all esxi hosts has to be in the same switch/vlan for connectivity, right now they are connected to an standard switch and i don't know how to connect my 3 esxi host to the same segment.... maybe a Distributed switch?

Regards

[piaroa]

No need for a vDS here. Just for the sake of trying, can you just run everything (Mgmt, vMotion, etc) on the same subnet and try? If it all works then you know it's something to do with the way you've set up your virtual lab on Workstation.

[mozer__]

I think you are right, if i set a private network on workstation for my vmotion nics i will have connectivity. The same for fault tolerance.

is like putting a real switch or a vlan on my esxi hosts, no routing device is needed... i think i got it :smileygrin:

You response was definitely helpful. I will post the update

Cheers!!

[mozer__]

Hi,

So, for anyone with the same problem, vmotion and Ha interfaces have to be in the same network, in a real environment that is accomplished with a real switch in a virtualized enviroment (like vmware workstation) a private LAN is needed

Regards