On 05/18/2016 01:53 AM, Edward Ahern wrote: The issue here isn't just one of IP addresses, but of connectivity. The virt-host has IP connectivity to the Undercloud through virbr0, which is the default libvirt bridge. The Undercloud nodes and Overcloud nodes are connected to brbm, which is isolated. I haven't tested this, but you might be able to add an external interface to the brbm bridge to connect the VMs to an external network: ovs-vsctl add-port brbm eth2 You would need to use a range of IP addresses for discovery and deployment that are unused on that network, and there can't be an external DHCP server on that network. Also, any clients on that network which are configured for DHCP or PXE boot will suddenly start booting into the discovery image, which might be really bad. Make sure that's what you want before testing it out. Another approach (also untested) might be to add an IP address on the brbm bridge on the virt-host, and use that IP as the default gateway for the ctlplane network. You could then configure IP forwarding on the virt-host and create a static route pointing to the ctlplane subnet via the virt-host. -- Dan Sneddon | Principal OpenStack Engineer dsneddon(a)redhat.com | redhat.com/openstack 650.254.4025 | dsneddon:irc @dxs:twitter

Hi, For this I also have the following question: On Fri, May 20, 2016 at 8:01 AM, Adam Young <ayoung(a)redhat.com&gt; wrote: I was wondering about the scope of the Quickstart documentation and the general TripleO documentation. What I think is the following: * Quickstart documentation should be simple and refer to the implementation and configuration of quickstart options * General TripleO would describe the planning, deployment, and troubleshooting of TripleO. It therefore might refer to the quickstart tool to perform a deployment... But as you can imagine, this will lead to having 'two truths' related to TripleO. Is the quickstart doing something so drastically different it needs it's own documentation? The quickstart tool itself describes how to access the console of the nodes, and also how to get to the dashboard. But this is because it now deals with a virtualized environment. More advanced how-tos now end up being captured in blogposts in several different places. How could this be improved? regards, Gerard -- Gerard Braad F/OSS & IT Consultant http://gbraad.nl