Proactive Tests - Multicast performance test

[AlexanderLiucka]

When you execute Proactive Tests - Multicast performance test, the test goes thru the Management VMkernel adapter instead VSAN VMkernel adapter, which apparently is an BUG.

the second strange think is on the generating interface I see only 233 Mbits per second and on the all receiving interfaces (6 interfaces, the VSAN cluster have 7 Hosts) I see around 230 Mbits per second but the end result is 51.09 MBs? Maybe this is BUG too?

[zdickinson]

Good morning, this has been covered here:  Why proactive multicast performance test fails when health check passes?  Thank you, Zach.

[AlexanderLiucka]

Hi zdickinson,

I have read the covered info about the problem, but there no one mentions what exactly is the problem, only that this is a BUG.

also when we know that the multicast test is running on the wrong VMkernel adapter those this mean that the result of the measurement also have to be wrong or to be correct?

[zdickinson]

I believe it's a mystery to us customers at the moment.  Perhaps Duncan, Cormac, or support could interject.  Thank you, Zach.

[CHogan]

This was highlighted in the VSAN 6.1 Release Notes as a known issue.

VMware Virtual SAN 6.1 Release Notes

• Multicast performance test of Virtual SAN health check does not run on Virtual SAN network
  In some cases, depending on the routing configuration of ESXi hosts, the network multicast performance test does not run on the Virtual SAN network. 

  Workaround: Use the Virtual SAN network as the only network setting for the ESXi hosts, and conduct the network multicast performance test based on this configuration.

  If ESXi hosts have multiple network settings, you also can follow the steps listed in this example. Assume that Virtual SAN runs on the 192.168.0.0 network.

  1. Bind the multicast group address to this network on each host:

     $ esxcli network ip route ipv4 add -n 224.2.3.4/32 -g 192.168.0.0?

  2. Check the routing table: $ esxcli network ip route ipv4 list
     default      0.0.0.0          10.160.63.253  vmk0       DHCP
     10.160.32.0  255.255.224.0    0.0.0.0        vmk0       MANUAL
     192.168.0.0  255.255.255.0    0.0.0.0        vmk3       MANUAL
     224.2.3.4    255.255.255.255  192.168.0.0    vmk3       MANUAL
  3. Run the proactive multicast network performance test, and check the result.
     
     
     
  4. After the test is complete, recover the routing table: $ esxcli network ip route ipv4 remove -n 224.2.3.4/32 -g 192.168.0.0

Pretty sure this also made it into a KB. Give this a go, see if it allows the multicast performance test to run.