Congratulations for the work Javi Roman !! It seems to me a very interesting tool to accelerate and test the work of the contributors. Personally, I would leave HDFS installed as a service to make management and maintenance easier. For demonstrations and tests, I think it would be interesting to put a spark-shell and update the README with some map-reduce example. What do you think?
On 9 April 2018 at 19:34, Javi Roman <jroman.espi...@gmail.com> wrote: > Hi! > > I've just created a PR [1] with the fist version of DC/OS (tested > v1.11.0) for Myriad development/deployment at DC/OS. > > The cluster created by default is something like that: > > $ vagrant status > Current machine states: > > bt running (libvirt) > m1 running (libvirt) > a1 running (libvirt) > a2 running (libvirt) > a3 running (libvirt) > a4 running (libvirt) > p1 running (libvirt) > > The bootstrap machine (with DC/OS CLI installed). One master node, 4 > private agents, and one public. > > I haven't tested Myraid yet, because we have different options for it: > > - Deploy HDFS as a system service. > - Deploy HDFS using Mesosphere Universe (we have to tune this one, the > requirements don't fit in this small Vagrant cluster). > - Deploy Myriad by means of Marathon and Docker. > > We have to test different options for this deployment. One interesting > idea would be create a Myriad Universe with packages for HDFS tuned > for this Vagrant, with upstream Apache Hadoop, and packages for > Myriad. This could be a good point for demos. > > > [1] https://github.com/apache/incubator-myriad/pull/108 > -- > Javi Roman > > Twitter: @javiromanrh > GitHub: github.com/javiroman > Linkedin: es.linkedin.com/in/javiroman > Big Data Blog: dataintensive.info >
