Traffic behavior in ESXi hosts directly connect via cable through vmnic

search cancel

book

calendar_today

VMware vSphere ESXi

This article provides understanding when using vmnics that are connected directly through physical cables and the corresponding vmnics failover.

From the above description, vmnic0/vmnic1 on hosts ESXI1 and ESXI2 are connected directly to each other.
vmnics are configured under the failover configuration “active-active” with load balancing policy set as Virtual Port ID.
For connectivity, VM1 (active -> vmnic0 ) from ESXi1 is trying to connect VM2 (active -> vmnic1) on ESXi2, but is failing to do so.
Performing trace packet-captures on ESXi2 shows the packets getting dropped by the virtual switch.
Login into the ESXi host as user root via putty/ssh

The VM/vmk from ESXi1 was sending the traffic over vmnic0 to ESXi2. However, the VM on ESXi2 was actively using vmnic1 and not vmnic0 for connectivity; thus the packets were being dropped on ESXi2 vSwitch.
Virtual switches do not have the mechanism to forward the traffic for a VM/vmk which is expected on one vmnic switchport to another vmnic’s switchport, unless they are configured under LACP/etherchannel, hence they are dropped on the switch.
If the vmnics are configured under LACP/etherchannel configuration, then based on the Load Balancing the traffic can be expected across the vmnics.
But, if there is no LACP/etherchannel, the vmnics should be configured under active-standby mode.

For these specific vmnic configurations, the vmnics on the virtual switch/portgroups should be configured under active-standby configurations.

If due to adapter issues a vmnic on one host goes down, on the other host the connectivity using the corresponding vmnic will be reported down as well.
“Network failure detection” will allow the vmnics to failover to the next available vmnic.

Output from trace capture:

thumb_up Yes

thumb_down No