I would like share with you the strategy that I'm following for a
powerful local development environment for Myriad.
The idea is to create an easy-to-use environment for new contributors,
setting up a Mesos cluster, or DCOS cluster (with hadoop running which
is a tedious task). I guess automate this task is a key point for
accelerating the development and focusing in the Myriad hacking.
In this first task [1] I am working on Vagrant libvirt support for a
plain Mesos cluster. However the roadmap for this environment would
be:
1. vagrant-libvirt based cluster for Mesos, provisioned with Ansible
(already working in my environment).
2. vagrant-libvirt based cluster for DC/OS provisioned with Ansible:
Easier environment for testing Myriad with Marathon, Mesos-DNS, VIPS,
.... and so on.
3. vagrant-virtualbox based cluster (maintain the original environment
used right now).
The layout would be something like this:
vagrant/
├── libvirt
│ ├── dcos
│ │ └── Vagrantfile
│ └── mesos
│ └── Vagrantfile
└── virtualbox
└── Vagrantfile
In a second stage, the final scenario would be, only one folder with
one "smart" Vagrantfile for supporting Mesos and DC/OS for Linux and
OSX, the tool will be able of selecting the suitable virtualization
technology by user configuration.
This is the idea.
[1] https://issues.apache.org/jira/browse/MYRIAD-257
--
Javi Roman
Twitter: @javiromanrh
GitHub: github.com/javiroman
Linkedin: es.linkedin.com/in/javiroman
Big Data Blog: dataintensive.info
