[jira] [Comment Edited] (BIGTOP-1547) Confirm tachyon functioning in vagrant recipes.

["jay vyas (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/BIGTOP-1547?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14224971#comment-14224971 ] 

jay vyas edited comment on BIGTOP-1547 at 12/2/14 4:37 PM:
-----------------------------------------------------------

Outline of this task..

First, setup a vagrant environment w/ tachyon added as yum repo to it:

- To test, you must build tachyon locally and add as a yum repo in the VM, that can be done following 
 - {{vagrant up}} to create a vm from vm/bigtop-deploy/vagrant-puppet, then
 - {{gradlew tacyhon-yum}} to create a yum repo for the tachyon package from source
then vagrant ssh
Now, ON THE VM:
 - run {{sudo yum-config-manager --add-repo file:///bigtop-home/output/ }} . 

Now: run {{puppet apply -d --confdir=/vagrant --modulepath=/bigtop-home/bigtop-deploy/puppet/modules /bigtop-home/bigtop-deploy/puppet/manifests/site.pp}} and  
- Confirm that puppet is properly and reproducibly setting up *slave* to properly point to *master* so no manual config is required.
- Add a "tachyon" option to the vagrant recipe, which , when turned on, allows the in memory cache *above the* running HCFS.

Please add questions in this thread, since *tachyon* is new, there may be some good insight we can discuss around preferred testing scenario.


was (Author: jayunit100):
Outline of this task..

First, setup a vagrant environment w/ tachyon added as yum repo to it:

- To test, you must build tachyon locally and add as a yum repo in the VM, that can be done following 
 - {{vagrant up}} to create a vm from vm/bigtop-deploy/vagrant-puppet, then
 - {{gradlew tacyhon-yum}} to create a yum repo for the tachyon package from source
then vagrant ssh
Now, ON THE VM:
 - run {{sudo yum-config-manager --add-repo file:///bigtop-home/output/ }} . 

Now: run {{puppet apply ...}} and  
- Confirm that puppet is properly and reproducibly setting up *slave* to properly point to *master* so no manual config is required.
- Add a "tachyon" option to the vagrant recipe, which , when turned on, allows the in memory cache *above the* running HCFS.

Please add questions in this thread, since *tachyon* is new, there may be some good insight we can discuss around preferred testing scenario.

> Confirm tachyon functioning in vagrant recipes.
> -----------------------------------------------
>
>                 Key: BIGTOP-1547
>                 URL: https://issues.apache.org/jira/browse/BIGTOP-1547
>             Project: Bigtop
>          Issue Type: Improvement
>          Components: vm
>    Affects Versions: 1.0.0
>            Reporter: jay vyas
>             Fix For: 1.0.0
>
>
> We have puppet recipes which install and setup tachyon at a basic level. 
> Lets *wire them into HDFS* directly, so that we can *run our smoke tests* against a hadoop cluster which has *tachyon* enabled and installed, by simply calling our favorite {{vagrant up}} command and watching the show unfold for us.
> [~evans_ye] CC on this as you expressed interest in learning more about tacyhon .  [~dcapwell] [~hsaputra] and [~chenh] might have feedback to questions you have on the tachyon deployment architecture itself. 



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)