Skip to content →

Month: November 2016

TripleO Overcloud deploy failed? Much sadness. Debugging your isolated network configuration

In setting up my home lab, Dr. Octagon I’ve been really stubbing my toe here and there getting the heat templates for network isolation just right. I still haven’t come exactly to the conclusions that I need to come to, but in the process I’ve gained a better understanding of the process for debugging what’s wrong with the templates in the process.

Comments closed

TripleO: Using the fake_pxe driver with Ironic

I’ve been working on testing things with TripleO and normally I use TripleO Quickstart to spin things up in a virtual environment.
Often when doing NFV work though, you need things that can’t be used in a virtual environment (such as DPDK, SR-IOV, etc) so you need some baremetal nodes.
In my home lab environment though, I don’t have the luxury of IPMI, so I need to make use of the fake_pxe driver in Ironic, which allows for standard PXE control, but requires you to deal with powering on and off the machines manually.

Comments closed

OpenShift-on-OpenStack – Creating a cluster with openshift-ansible

We’re going to try to use the openshift-ansible to spin up an OpenShift cluster. In the “easy mode” article about OpenShift, we created just a single play node, and also a whopping single point of failure. So let’s build up on that. We’ll take this to the point where we can see the pods that are running, and see where the dashboard is, and in a further article explore how to set up the networking and at least make it a little more real-world-useable.

Comments closed

openshift-ansible – Bring your own inventory.

I’m working on getting an OpenShift cluster up and running, and in the while of trying to do it properly with openshift-ansible’s openstack method (which uses heat templates, as opposed to us building it all by hand here), I was trouble shooting an issue (my clocks were…. quite wrong, causing me much grief with creating proper certs for HTTPS necessary for etcd)… And I wound up going and doing the “create your own inventory” method.

Comments closed

Creating a Custom Centos Cloud Image (for OpenStack)

So I’m working on getting openshift-ansible to spin up on top of OpenStack. And I’ve run into a few snags, one of which requires that I need to make a custom cloud image to run because the CentOS Generic Cloud image doesn’t have NetworkManager install/enabled, and… Even worse but maybe a story for another day is that it doesn’t support a Centos Atomic Host. Another approach would be to try to improve the Ansible script in openshift-ansible, however, this slices the Gordian Knot for now, and we’ll come back to that another time (as there’s plenty else to let ansible-openshift know about as you’ll find out in an upcoming article).

Comments closed