Hi,
I'm trying to get VXLAN working with host profiles on a group of stateless hosts that I am setting up for vCloud Director use. I've successfully gotten the hosts to auto deploy with the right agents installed (esx-vxlan, etc). I'm able to configure the VXLAN manually on the stateless hosts and they show up with green checks in the Preparation tab in vShield Manager.
(My hosts are running 5.1U1 but I saw this behavior with 5.1, too. The vShield Manager is 5.1.2a and the VXLAN VIB is pulled from that.)
When I reboot the hosts, I see the following tasks happen as they come back up in maintenance mode:
- Disconnect host - succeeds
- Reconnect host - succeeds
- Enable agent - succeeds. I believe that this is enabling of the VXLAN agent.
- Scan - succeeds. Another related to VXLAN.
- Add virtual NIC - fails. I believe the failure is because the host profile hasn't been applied yet, so there is no vDS for the VXLAN agent that was enabled to modify. Normally, this would add the vmk for the VXLAN portgroup on the vDS.
At this point, sometimes more processing of the host occurs. Sometimes not. If more happens, it's the following:
6. Apply host configuration - sometimes succeeds.
7. Exit maintenance mode - succeeds.
8. Add virtual NIC - succeeds. This causes a 2nd vmk to be added to the portgroup for VXLAN.
Once the process is completed, if the host makes it out of maintenance mode, then it usually is running fine except for the VXLAN configuration which is all kinds of borked up. I can try to apply the host profile again, but when I exit maintenance mode after that, the "Add virtual NIC" task gets kicked off and a 2nd vmk gets applied.
I can go through and manually clear things up, but that seems to defeat the purpose of the host profiles.
It seems like there are two problems:
- The VXLAN agent enables too early. It shouldn't kick off until after the profile is applied.
- The VXLAN agent doesn't seem to recognize that a vmk already exists in the portgroup after the host profile is applied and tries to create its own, which doesn't have the right configuration.
I'm not sure how to proceed here and get this working.
Has anyone encountered this or have any suggestions?
Thanks,
Jason