The team that I am working with right are very much agile and we’re doing quite a bit of outside in development of the repository that we’re building. We’re mostly adopting a behaviour driven development with a touch of test driven development. As a result we’re very much in favour of testing things out as much as we can and using the same environments to develop against. As previously mentioned before I had originally been using puppet and vagrant to build up the development harness and experiment with tools/services that we might want to use for our system.
At somepoint I came across ansible and not long after discovering it, I migrated a large chunk of the development and test systems to using ansible. I’ve even cooked up one or two ansible modules as a result.
As a result of adopting ansible for building up our test and development infrastructure, I’ve collected the relevant playbooks and roles that a person might want for deploying all the bits and pieces needed to roll out a hydra-head. See https://github.com/jcftang/ansible-hydra, I have a set of roles and a few example playbooks on setting up at least
- Tomcat (from the base repositories of RHEL6/Centos6/ScientificLinux6)
- Fedora-Commons (the same version as is in the hydra-jetty repo)
- Apache SOLR (the same version as is in the hydra-jetty repo)
- Ruby (via RVM) in a user directory
- Passenger with the installed version of Ruby
- MySQL (from the base repositories of RHEL6/Centos6/ScientificLinux6)
- Redis
The configurations aren’t quite production ready yet as they do require some more work in setting up Fedora-Commons and SOLR the way we want. The configurations are however fairly realistic and are daily use for doing test deployments of our hydra-head (RoR application) or experimenting with additional tools, configurations and systems such as ceph - we’re using the radosgw to provide a realistic and local S3 service.
So far the configurations need polishing off and another playbook needs to be created for continuous deployment of our hydra-head.