corcovadoming / dockerambari

Modifed ambari on docker for data scientist

DockerAmbari

Introduction:

This repository is forked from https://github.com/sequenceiq/docker-ambari with modified blueprint, which brings up a cluster for data scientist in Docker with minimum CPU and Memory usage.

Request:

• Docker
• Bash-based shell
• This repository is only tested on Linux

Install:

1. $ source helper, to have amb prefix functions

2. $ amb-start-cluster <number-of-cluster> startup a cluster with consul

3. Launch ambari-shell by amb-shell

4. blueprint add --file data_scientist.bp to add the blueprint for data scientist pre-installed software

5. cluster autoassign to deploy to your cluster

6. cluster create --exitOnFinish true Done!

Nginx

In order to dynamically mapping the port from container to host, modified nginx which forked from https://hub.docker.com/r/jlordiales/nginx-consul/ is used

1. docker build -t <nginx-image-name> . to build it up

2. Launch the nginx container by

$ docker run \
  --name nginx \
  -d \
  -p 8080:8080 -p 9995:9995 -p 9996:9996 \
  --v `pwd`/service.ctmpl:/templates/service.ctmpl \
  --dns $(docker inspect --format="{{.NetworkSettings.IPAddress}}" amb-consul) \
  --dns 172.17.0.1 \
  --dns 8.8.8.8 \
  --dns-search service.consul \
  <nginx-image-name>

which forwards port 9995 for zeppelin notebook and 8080 for ambari dashboard