[
https://issues.apache.org/jira/browse/BIGTOP-1178?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13911748#comment-13911748
]
Evans Ye commented on BIGTOP-1178:
----------------------------------
Good News!
I've worked out the first version of the patch.
In this patch I use [Vagrant Host
Manager|https://github.com/smdahlen/vagrant-hostmanager] to manage VMs'
/etc/hosts to ensure that every one in cluster know each other.
A simple shell wrapper {{startup.sh}} is used to spinning up VM(s) and specify
which mode of deployment is going to be chose, could be standalone mode or
cluster mode.
I've also wrote down steps to use in a README.md which shipped with the patch.
Please give me any suggestion or feedback when something pop up in your mind,
thanks
> Clusterize the puppetized vagrant deployer.
> -------------------------------------------
>
> Key: BIGTOP-1178
> URL: https://issues.apache.org/jira/browse/BIGTOP-1178
> Project: Bigtop
> Issue Type: Improvement
> Components: Deployment
> Reporter: jay vyas
> Priority: Minor
>
> Now that https://issues.apache.org/jira/browse/BIGTOP-1171 is nearing
> completion (puppetization of vagrant), we can move towards setting up a real
> hadoop "cluster".
> The goal here will be to allow us to do more with vagrant than just spinning
> up a single VM. To scale the vagrantfile, we can do something like :
> {noformat}
> if ARGV[1] == 'cluster'
> cluster = true
> else
> cluster = false
> end
> {noformat}
> I think the Vagrantfile can look something like this:
> {noformat}
> Vagrant.configure(VAGRANT_API_VERSION) do |config|
> # Head node
> config.vm.define :bigtop1 do |bigtop1|
>
> # If "cluster" then add a bunch of slaves also
> if cluster == true
> # Slave nodes ....
> config.vm.define :bigtop2 do |bigtop2|
> config.vm.define :bigtop3 do |bigtop3|
> config.vm.define :bigtop4 do |bigtop4|
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
