We are preparing to replace some of the network switches that our VMware environment runs on. When I ran a test with a production host running a couple of test vm’s I saw some unexpected results. Apparently I have an incorrect configuration or an incomplete understanding of what to expect in a fail over mode.
We are running ESXi 5.0.0, build 821926. The vSphere Client is version 5.0.0 Build 804277. Management and Vmotion traffic is carried on the same vlan. I understand this isn’t Best Practice but it is what I have to work with. There are 2 NIC’s (vmnic0 and vmnic4) that carry this traffic to 2 switches.
Vmnic0 & vmnic4 support a vSphere Standard Switch, vSwitch0. The nics are configured active/active. There are 3 port groups configured on the switch. One is a VMkernel Port Group named “Management Network”. Management traffic is enabled for this port group and vmotion traffic is not. Vmnic0 is active for the port group and vmnic4 is standby.
The vMotion port group is named ”vMotion”. Vmnic4 is active for this port group and vmnic0 is standby.
The last port group is named “VM Network” and is a Virtual Machine port group. Both vmnic0 & vmnic4 are active for this port group.
If it matters, iSCSI traffic is routed to a different physical switch.
We have configured and installed the replacement switches. The test was to one at a time, move the vmnic0 and vmnic4 connections from the old switch to the new. When I disconnected the active management connection (vmnic0) I was expecting management traffic to failover to the standby connection (vmnic4). What I observed was the vSphere lost connection with the host. Somewhat surprised at this I attempted to log in directly to the host from the vSphere client but was unable to connect. If I restored the connection to either the original switch or the new switch, vSphere immediately showed the host connected. I repeted the test after enabling Management Traffic on the vMotion port group. I saw the same result but I didn't boot the host however.
Is my expectation not correct in expecting management traffic to failover to the standby nic or is the configuration incorrect for failover to occur?
TIA