I realize this is a rather old thread, but for someone (like me) who is still having this issue, this is what I did. I am using Mirantis OpenStack/Fuel 9, on hardware/KVM Ubuntu Trusty 14.x:

The only change I can see that made any difference was to go into the controller node(s) and change the following vnc section (near the end of the file /etc/noca/nova.conf):

[vnc] novncproxy_host=0.0.0.0 novncproxy_port=6080 novncproxy_base_url=http://192.168.0.3:6080/vnc_auto.html

192.168.0.3 is my HA IP, the big change was (I think) adding the host and port. Then restarted using:

/etc/init.d/neutron-server restart

need to check the firewall

Even-though I had disabled the firewall between all nodes

after hours of looking around, tcpdump to the rescue

on the controller node

tcpdump -nni eno1 port 5900

replace eno1 with the management interface , try to launch your vnc-console, you should see a single line computenode:5900

13:55:43.880155 IP 10.0.70.70.45261 > 10.0.70.93.5900: Flags [S], seq 1495139310, win 14600, options [mss 1460,sackOK,TS val 12232244 ecr 0,nop,wscale 7], length 0

Go to the computer node and issue iptables -L you may be surprise as I was to see it was running

Again issues iptables -F to drop all tables. Reload your dashboard and launch your vnc-console again

... and smile (if it was the same problem as mine)