[jira] [Comment Edited] (BIGTOP-2615) Provide a tool to build pseudo cluster docker images

["Evans Ye (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/BIGTOP-2615?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15719554#comment-15719554 ] 

Evans Ye edited comment on BIGTOP-2615 at 12/4/16 8:19 AM:
-----------------------------------------------------------

Currently systemd based deployment are not supported because there's no way to get CAP_SYS_ADMIN during image build time. I've noted that in README.md.

OTOH, I've drafted a jenkins job for producing Hadoocker images:
https://ci.bigtop.apache.org/view/Docker/job/Hadoocker/

And pushed one testing image for test:
{code}
docker run -ti --name mystack --rm -p 8088:8088 bigtop/hadoocker:centos-6_hadoop_yarn_spark
docker exec -ti mystack /usr/lib/spark/bin/spark-shell --master yarn
{code}


was (Author: evans_ye):
Currently systemd based deployment are not supported because there's no way to get CAP_SYS_ADMIN during image build time. I've noted that in README.md.

OTOH, I've drafted a jenkins job for producing Hadoocker images:
https://ci.bigtop.apache.org/view/Docker/job/Hadoocker/

And pushed one testing image here:
bigtop/hadoocker:centos-6_hadoop_yarn_spark


> Provide a tool to build pseudo cluster docker images
> ----------------------------------------------------
>
>                 Key: BIGTOP-2615
>                 URL: https://issues.apache.org/jira/browse/BIGTOP-2615
>             Project: Bigtop
>          Issue Type: New Feature
>          Components: docker
>    Affects Versions: 1.1.0
>            Reporter: Evans Ye
>            Assignee: Evans Ye
>         Attachments: Hadoocker Design Concept.pdf
>
>
> This JIRA implements a tool named Hadoocker, which is to build docker images that have specified hadoop components provisioned inside. During image build time, we execute puppet apply to install packages and configuration. After images are built, components can be brought up quickly by executing puppet apply again. Usually it takes about 80~180 seconds, depends on the complexity of components.
> This work adopt exactly the same idea introduced by [~cos] in BIGTOP-2296, but goes further to implement a tool that supports different OS, components, or even a customized site.yaml config file.
> Use cases for Bigtop users:
> * Quick POC environment
> * Being leveraged in CI/CD pipeline
> Use cases for Bigtop:
> * Image generation CI jobs can be treated as our integration tests.



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