commit	fdeecf81f5512bffff09f4c48e62ea138b01fe7c	[log] [tgz]
author	Mike Walch <mwalch@apache.org>	Mon Sep 11 11:05:07 2017 -0400
committer	GitHub <noreply@github.com>	Mon Sep 11 11:05:07 2017 -0400
tree	4b97af50b3cd55eebf973207b80d6ab34a6efadb
parent	23119dd749569582e6e9d616b0463fb78b52fac0 [diff]

tree: 4b97af50b3cd55eebf973207b80d6ab34a6efadb

README.md

Fluo Docker Image

Obtain the Docker image

To obtain the docker image created by this project, you can either pull it from DockerHub at mikewalch/fluo or build it yourself. To pull the image from DockerHub, run the command below:

sudo docker pull mikewalch/fluo

While it is easier to pull from DockerHub, the image will default to the software versions below:

Software	Version
Fluo	1.2.0-SNAPSHOT
Accumulo	1.8.1
Hadoop	2.7.3
Zookeeper	3.4.9

If these versions do not match what is running on your cluster, you should consider building your own image with matching versions. However, Fluo must be 1.2+. Below are instructions for building an image:

Clone the Fluo docker repo

 git clone git@github.com:mikewalch/fluo-docker.git

Until Fluo 1.2 is released, build a Fluo tarball distribution and copy it to the root directory of the repo.

 git clone git@github.com:apache/fluo.git
 cd fluo/
 mvn clean package
 cp modules/distribution/target/fluo-1.2.0-SNAPSHOT-bin.tar.gz /path/to/fluo-docker/

Build the default Fluo docker image using the command below.

 cd /path/to/fluo-docker
 docker build -t fluo .

Or build the Fluo docker image with specific versions of Hadoop, Zookeeper, etc using the command below:

 docker build --build-arg ZOOKEEPER_VERSION=3.4.8 --build-arg ACCUMULO_VERSION=1.7.3 --build-arg HADOOP_VERSION=2.7.0 -t fluo .

Image basics

The entrypoint for the Fluo docker image is the fluo script. While the primary use case for this image is to start an oracle or worker, you can run other commands in the fluo script to test out the image:

# No arguments prints Fluo command usage
docker run mikewalch/fluo
# Print Fluo version
docker run mikewalch/fluo version
# Print Fluo classpath
docker run mikewalch/fluo classpath

Run Fluo Applications using Docker

Before starting a Fluo oracle and worker using the Fluo Docker image, initialize your Fluo application. Next, choose a method below to run the oracle and worker(s) of your Fluo application. In the examples below, the Fluo application is named myapp and was initialized using a Zookeeper node on zkhost.

Docker engine

Use the docker command to start local docker containers.

Start a Fluo oracle

 docker run -d --network="host" mikewalch/fluo oracle -a myapp -o fluo.connection.zookeepers=zkhost/fluo

Start Fluo worker(s). Execute this command multiple times to start multiple workers.

 docker run -d --network="host" mikewalch/fluo worker -a myapp -o fluo.connection.zookeepers=zkhost/fluo

Marathon

Using the Marathon UI, you can create applications using JSON configuration.

The JSON below can be used to start a Fluo oracle.

{
  "id": "myapp-fluo-oracle",
  "cmd": "fluo oracle -a myapp -o fluo.connection.zookeepers=zkhost/fluo",
  "cpus": 1,
  "mem": 256,
  "disk": 0,
  "instances": 1,
  "container": {
    "docker": {
      "image": "mikewalch/fluo",
      "network": "HOST"
    },
    "type": "DOCKER"
  }
}

The JSON below can be used to start Fluo worker(s). Modify instances to start multiple workers.

{
  "id": "myapp-fluo-worker",
  "cmd": "fluo worker -a myapp -o fluo.connection.zookeepers=zkhost/fluo",
  "cpus": 1,
  "mem": 512,
  "disk": 0,
  "instances": 1,
  "container": {
    "docker": {
      "image": "mikewalch/fluo",
      "network": "HOST"
    },
    "type": "DOCKER"
  }
}