After VMotion, unable to ping VM's on other ESX server

[Chandrashekhar]

Hello everyone,

I have two ESX servers and some VM's on top of them. I am facing one strange problem from last few days. Whenever I do VMotion, I lost network connection to VMotioned VM. If I change the speed of physical NIC to 100MB/s, I am able to connect to VMotioned VM.

When I change the speed settings of physical NIC, "network " in settings changes automatically. Is this "network" information depends on NIC speed?

I had posted the same problems in communities but couldnt find any solution.

After VMotion, I am able to :

1) From VMotioned VM, ping ESX server on which VM is currently runnning.

2) From VMotioned VM, ping default gatteway of VMKernel.

3) From VMotioned VM, ping VMkernel IP address on which VM is running.

4) From VMotioned VM, ping VM's on same ESX server.

5) From VM's on other ESX server, ping VM's on different ESX server.

and unable to :

1) From VMotioned VM, ping other ESX server.

2) From VMotioned VM, ping VM's on other ESX server.

3) From other VM's cannot ping VMotioned VM.

I have kept same default gateway on both esx servers also enabled "notify switch" option.

Does anyone knows why this happens?

Please note that, if I change physical NIC settings to 1000mp/s, I can ping to and from VMotioned VM.

I do not have access to physical switch to make any changes to it.

[weinstein5]

Changing the speed of the physical NIC on the ESX server that the VM has been vmotioned allows for communication - I would take a look at the physical switch that NIC is connected to and insure the port's speed and duplez is fixed

If you find this or any other answer useful please consider awarding points by marking the answer correct or helpful

[Chandrashekhar]

Hi,

I do not have access rights to physical switch so cannot do anything about it. Can you please tell me where can I find more information about this physical NIC and its speed settings? Also, what can be the possiblereason behind this problem? Is it a bug in ESX software or what?

[weinstein5]

Information about the Physical NIC can be found under the ESX Host configuration tab it is where you changed the speed of the physical NIC - I really think the problem is with the physical network - either there is an issue with the switch configuration or the vswitch is not ocnnected to right network segment through the physical NIC assigned to it -

If you find this or any other answer useful please consider awarding points by marking the answer correct or helpful

[Chandrashekhar]

Thanks Weinstein,

I will check the physical network connections.

[speedyh]

i'm having the same issue with vmotion between 2 new HP DL380 G6 and SAN.

Both ESX servers are connected to gigabit switch, NIC are configured on 1000Mb Full duplex.

Changing speed on physical nic made the ESX unresponsive.

[dickybird]

yes it does on ESX like any physical windows/Unix servers.

The speed/duplex should be maching the netowork switch to avoid disconnections.

also in case if ESX servers try to use same speed/duplex and network switch settings across the ESX hosts in the farm alongwith same Vswitch settings on all ESX servers. e.g if Vswitch2 is Public on all ESX servers.

It should not be Vswitch2 public on ESX1 and Vswitch3 public on ESX2..thats wrong

[maartenn2]

We had the same problem.

Vmotion of a VM and no external network connections to that VM for 5 minutes.

In our case it was a HP Procurve 2510G-24 switch that did not refresch it's arp- table in time. The firmware Y_11_08 solved this problem.

Regards,

[tuomapet]

I can confirm the exact problem with HP Procurve 2510G-24 switches (old FW 11.01). Fortunately, since there was three switches for ESX servers alone, it was easy to update the firmware one at a time.