Strange behavior in ESXi for an network card with Npar.

Hi All

I cannot explain the following issue: Ping transmission fails for the configuration when vmnics from the same partition are added to the same vSwitch.

Hardware used:

- Network adapter: four-port Intel X710 with Dell firmware
- ESXi: 6.7 GA
- Npar partitions: four partitions, two vmnics per each partition (one vmnic master and one vmnic slave)
- Two servers Dell R730 connected by an external physical switch. NPAR enabled only for one server (SUT).

System Under Test (SUT):

Both physical adapters are active. Vmnic16 - master and vmnic20 - slave, both of the same partition.

Basically,  I pinged these three vmkernel adapters from other server. Ping to the first one (vmk3) failed.  Is it a feature or a bug?

Below, I have attached ESXi console screenshot to make my setup more understandable: 

1. Partition pairs. We can see two partition pairs which are up: the green and yellow one 

2. ESXTOP view of the testing setup:

3. Ping test results:

If I change the setup and replace a slave vmnic (vmnic20 - green partition) to a slave vmnic from the second partition (vmnic21 - yellow partition) then all ping tests are passed.