Ask Your Question

Revision history [back]

click to hide/show revision 1
initial version

Instance IP address not reachable

I have installed OpenStack Liberty with 1 controller and 1 compute node. The two nodes have 2 NICs: the management interface is on 10.0.0.x and the second NIC is unnumbered per the environment setup instructions.

Because of the limited IP address ranges I have available to me, I want the instance public IP's to be on the 10.0.0.x subnet, which is the same subnet as the management network.

When everything was built, I used the Horizon interface to create and launch an instance using the cirros image that is installed as part of the deployment guide. Horizon tells me that my instance is up and running and has an IP address of 10.0.0.202, which is in the DHCP range I created.

However, I can't ping this instance from either the management node nor the compute node, nor from another host on the 10.0.0.x subnet. Additionally, the Console tab in Horizon shows "Server disconnected (code: 1006)."

If I edit /etc/neutron/plugins/ml2/linuxbridge_agent.ini to use the management NIC for the public physical_interface_mapping on the compute and controller nodes, and then reboot both nodes and reboot the instance, the same thing happens.

My virtual router is on IP address 10.0.0.201, and I can ping that virtual interface. I have my instance (10.0.0.202) connected to the "Public" network, which is also connected to the router (10.0.0.201) per the Network Topology view.

How do I troubleshoot this?