sequenceiq/ambari-shell repository overview

⁠Ambari and Ambari Shell in Docker

This docker image aims to get you started with Apache Ambari⁠ and Ambari Shell⁠

⁠TL;DR

for the impatient, here is how you bring up a single-node ambari 'cluster', and use a blueprint to install services

docker run -d -p 8080 -h amb0.mycorp.kom --name ambari-singlenode sequenceiq/ambari --tag ambari-server=true
docker run -e BLUEPRINT=single-node-hdfs-yarn --link ambari-singlenode:ambariserver -t --rm --entrypoint /bin/sh sequenceiq/ambari-shell -c /tmp/install-cluster.sh

or if you want to have a one-liner which fits into a twitter message:

curl -LOs j.mp/ambari-singlenode && . ambari-singlenode

⁠Requirement

The only software you need is Docker⁠. If you haven't installed it yet, you need to get out from under that rock you've been living, and read the installation⁠ guide.

⁠Get a running Ambari Server url

You need a running Ambari Server where you can connect with ambari shell. You either:

• Have an already running Ambari Server, for example HDP⁠ sandbox.
• You need an Ambari Server sandbox you can easily play with, and than discard.

⁠Option-A: Ambari Server on docker

You can start a single-node ambari "cluster", with agent and server running in the same container:

docker run -d -p 8080 -h amb0.mycorp.kom --name amb0 sequenceiq/ambari --tag ambari-server=true

⁠Option-B: BYOA (Bring You Own Ambari)

You just replace <HOSTNAME> with the actual Ambari Server's hostname or IP.

docker run -it --rm  sequenceiq/ambari-shell --ambari.host=<HOSTNAME> --ambari.port=<PORT>

⁠Wait for ambari server

Ambari server needs some time to start up. It will respond with RUNNING on the REST endpoint: /api/v1/check. You either wait for 10-20 sec, or if you prefer to watch a growing dotted line in your terminal:

⁠Option-A: Ambari Server on docker

we have a script prepared inside of the docker image

docker run --link amb0:ambariserver -e EXPECTED_HOST_COUNT=1 -it --rm --entrypoint /bin/sh sequenceiq/ambari-shell -c /tmp/wait-for-host-number.sh 

Notable parameters

• --entrypoint /bin/sh the default entrypoint is to run the ambari-shell java process, it hast to be changed, read more best practices⁠
• -e EXPECTED_HOST_COUNT=1 sets en env variable for the script, it can be omitted if you are good with the default 1.
• --link amb0:ambariserver this will link the container named amb0 to the container running the script. Docker will prepare a couple of env variable with information about the linked amb0. For example: AMBARISERVER_PORT_8080_TCP_ADDR will point to Ambari Server listen address. This way you don't have to explicitly specify the AMBARI_HOST env variable. Which is used in BYOA section.

⁠option-B: BYOA

if you want to wait for a not dockerized ambari server

docker run -e AMBARI_HOST=172.19.0.45 -e EXPECTED_HOST_COUNT=2 -it --rm --entrypoint /bin/sh sequenceiq/ambari-shell -c /tmp/wait-for-host-number.sh

⁠Automated cluster creation

DevOps first please! So the description of the automated cluster installation goes like:

docker run -e BLUEPRINT=single-node-hdfs-yarn --link amb0:ambariserver -it --rm --entrypoint /bin/sh sequenceiq/ambari-shell -c /tmp/install-cluster.sh

⁠Start ambari-shell

To run ambari shell, you just start a yet another docker container:

AMBARI_IP=$(docker inspect --format "{{.NetworkSettings.IPAddress}}" ambari-singlenode)
docker run -it --rm sequenceiq/ambari-shell --ambari.host=$AMBARI_IP

Once the shell is started you will see the following banner

    _                _                   _  ____   _            _  _
   / \    _ __ ___  | |__    __ _  _ __ (_)/ ___| | |__    ___ | || |
  / _ \  | '_ ` _ \ | '_ \  / _` || '__|| |\___ \ | '_ \  / _ \| || |
 / ___ \ | | | | | || |_) || (_| || |   | | ___) || | | ||  __/| || |
/_/   \_\|_| |_| |_||_.__/  \__,_||_|   |_||____/ |_| |_| \___||_||_|

Welcome to Ambari Shell. For command and param completion press TAB, for assistance type 'hint'.
ambari-shell>

⁠Create a cluster

Now ambari shell is connected to the server and waits for you commands. The hint command can guide you through the process of creating a cluster, but here are the bare minimum:

blueprint defaults
blueprint list
cluster build --blueprint single-node-hdfs-yarn
cluster autoAssign
cluster create
tasks