Comment 9 for bug 1669468

Revision history for this message
Markus Zoeller (markus_z) (mzoeller) wrote :

Change https://review.openstack.org/#/c/444164/6 didn't solve it completely.
The VNC settings are not complete. The controller node (with compute service)
is set up correctly:

    [vnc]
    xvpvncproxy_host = 0.0.0.0
    novncproxy_host = 0.0.0.0
    vncserver_proxyclient_address = 158.69.78.118
    vncserver_listen = 0.0.0.0
    xvpvncproxy_base_url = http://158.69.78.118:6081/console
    novncproxy_base_url = http://158.69.78.118:6080/vnc_auto.html

But the compute node (subnode-2) is not:

    [vnc]
    xvpvncproxy_host = 0.0.0.0
    novncproxy_host = 0.0.0.0
    vncserver_proxyclient_address =
    vncserver_listen = 0.0.0.0
    xvpvncproxy_base_url = http://158.69.78.118:6081/console
    novncproxy_base_url = http://158.69.78.118:6080/vnc_auto.html

Note the missing IP address for "vncserver_proxyclient_address". This is
probably due to the wrong order of the variables in the "local.conf" of
the subnode:

    NOVA_VNC_ENABLED=true
    VNCSERVER_LISTEN=0.0.0.0
    VNCSERVER_PROXYCLIENT_ADDRESS=$HOST_IP

    HOST_IP=158.69.78.119

The "HOST_IP" has to be *before* the "VNCSERVER_PROXYCLIENT_ADDRESS".
Logs: http://logs.openstack.org/76/400876/31/check/gate-tempest-dsvm-neutron-multinode-full-ubuntu-xenial-nv/63d2bdd/logs/

The Logstash query: http://logstash.openstack.org/#/dashboard/file/logstash.json?from=7d&query=message:%5C%22AssertionError:%20True%20is%20not%20false%20:%20Token%20must%20be%20invalid%20because%20the%20connection%20closed.%5C%22