MiniFluo Setup

Below are instructions for running a test/development instance of Fluo called MiniFluo. MiniFluo runs its own Accumulo and Zookeeper but can only run on one machine. While MiniFluo is easy to set up and has all of the features of Fluo, it is not recommended for production use as all of its data is lost when its stopped.

If you looking to run Fluo in a production environment, read the production setup instructions.

Obtaining a distribution

Before you can install MiniFluo, you will need to obtain a distribution tarball. You can either download the latest release or build a distribution tarball by following these steps:

First, clone Fluo:

git clone
cd fluo/

Optionally, check out a stable tag (if you don't want to build a release from master):

git checkout 1.0.0-beta-1

Next, build a distribution which will be created in modules/distribution/target.

mvn package

Install Fluo distribution

When you have a distribution tarball built for your environment, follow these steps to install Fluo.

First, choose a directory with plenty of space and untar the distribution:

tar -xvzf fluo-1.0.0-beta-1-bin.tar.gz

Verify that your distribution has the same versions of Hadoop & Accumulo as your environment:

cd fluo-1.0.0-beta-1
ls lib/hadoop-* lib/accumulo-*

Next, copy the example configuration to the base of your configuration directory to create the default configuration for your Fluo install:

cp conf/examples/* conf/

The default configuration will be used as the base configuration for each new application. Therefore, you should modify for your environment. However, you should not configure any application settings (like observers).

As you are running a MiniFluo instance, most properties in do not need to be set by you. For example, all “Client properties” can be left unset as MiniFluo will start up its own Accumulo and Zookeeper. If you would rather have MiniFluo connect to an existing Accumulo instance, change to false and set all “Client properties”. If you have an existing Accumulo instance, you should also consider running a Fluo instance by following the production setup instructions.

MiniFluo command script

The MiniFluo command script is located at bin/mini-fluo of your Fluo installation. All MiniFluo commands are invoked by this script.

Modify and add the following to your ~/.bashrc if you want to be able to execute the mini-fluo script from any directory:

export PATH=/path/to/fluo-1.0.0-beta-1/bin:$PATH

Source your .bashrc for the changes to take effect and test the script

source ~/.bashrc

Running the script without any arguments prints a description of all commands.

Configure a MiniFluo application

You are now ready to configure a MiniFluo application. Use the command below to create the configuration necessary for a new application. Feel free to pick a different name (other than myapp) for your application:

mini-fluo new myapp

This command will create a directory for your application at apps/myapp of your Fluo install which will contain a conf and lib.

The apps/myapp/conf directory contains a copy of the from your default configuration. This should be configured for your application:

vim apps/myapp/

When configuring the observer section in, you can configure your application to run the phrasecount example if you have not developed your own application. See the phrasecount example for instructions. You can also choose not to configure any observers but MiniFluo will be idle when started.

The apps/myapp/lib directory should contain any observer jars for your application. If you configured for observers, copy any jars containing these observer classes this directory.

Run a MiniFluo application

A MiniFluo application can be started with the following command:

mini-fluo start myapp

The application will output its logs to app/myapp/logs/ directory of your Fluo installation.

MiniFluo starts its own cluster of Accumulo and Zookeeper. All data for this cluster is written by default to app/myapp/mini/ directory of your Fluo installation but this can be configured in Logs for underlying cluster can be viewed in mini/logs.

Due to MiniFluo starting its own cluster, it generates and writes a file to its data directory. This file can be used by Fluo clients to connect to the MiniFluo instance.

When you have data in your fluo instance, you can view it using the command mini-fluo scan myapp. Pipe the output to less using the command mini-fluo scan myapp | less if you want to page through the data.

Stop a MiniFluo application

A MiniFluo application can be stopped by running the following command:

mini-fluo stop myapp

If MiniFluo started its own Accumulo cluster, it will also remove the mini/ data directory and any data in your instance will be lost.