Day 3, last day, today only has 2 sessions that I was going to attend. I started out by going to an "Ansible best practices Working Session" by Michael Scherer. The goal was to cover Ansible basics, best practices, and how they apply those to the Fedora Infrastructure. One example he used was checking the checksums on files before you replace them and restart services with Ansible. In particular, ssh config files. You can imagine if you restart ssh on your clusters across datacenters and you break ssh... no more Ansible. Another best practice is to leverage the pkg module which can determine which package manager is being used by the host and adjust accordingly. The third best practice was to be careful on how you assign variables. Try to use local when possible.The Fedora Infrastructure team keeps thier ansible code
here Michael spent quite a bit of time covering the organization of their playbooks, roles, groups, tasks, handlers. A couple of tools that might be helpful:
Michael also walked through a playbook in detail with the upstream maintainer of the mailman package. There were quite a few best practices thrown around for that one.
After the session Michael and I had a chance to sit down and talk about the OpenShift Origin deployment for the cloud working group. Some decisions need to be made:
- Deploy Origin containers on Fedora Atomic or Fedora 24
- Set some expectations that this may be redeployed while we are learning?
- Bare metal or OpenStack?
- Storage for registry? Swift (not in Fedora?) NFS (hope not)
- Architecture: What do we want to deploy? Doesn't have to be production quality.
So that's the great thing about coming to a conference like this. Get a chance to put some faces to names and talk about fun, important projects. Flock is now over and my overall impression is that this conference was run very well. Lots of activities, food was great, people were great, sessions were great. I'm looking forward to my next Flock.
No comments:
Post a Comment