VCENTER_ESX_HOST inconsistent produces novnc 1006 error


#1

Hello, I’ve detected eventual errors after launching vmware instances from one, when I try to connect to the console with noVNC, it reports code 1006. Verified it is not due to firewall rules in the hosts.
After investigating a little I’ve detected that the VCENTER_ESX_HOST variable in the “vCenter Information” section is wrong, and looking below in the “Monitoring” section is reported correctly. If I move the VM to the host reported in the “vCenter” section it works, but leaving as is the failure kept.


In previous capture you can see the inconsistence at both sections.
Is it a bug?

OpenNebula 5.6.1
VSA 6.7.0.20000


(Jeremy) #2

I’d like to bump this up in the list… I too am having the same issues.
It seems that when a VM is deployed it chooses a host based on the scheduler. After some time DRS moves the system around, and updates the VCENTER_ESX_HOST variable in the monitoring section. Looking at the novnc.log the connections are attempting to use the VCENTER_ESX_HOST from the “vCenter Information”, which didn’t update.
Therefore my vnc connection attempt is being made to the incorrect ESXi host.

Is anyone else having these issues? I’ve been scouring forums and configuration guides but haven’t seen a mention of these issues.


(Jeremy) #3

I spoke too soon:
There is a fix in 5.6.2

VCENTER_ESX_HOST fail with DRS in vCenter #2477

https://github.com/OpenNebula/one/issues/2477