BUILD THE AMI

1. From the ami_build directory, run ami_build.py
    * Use -h for a list of parameters you can supply. 
    * You must at the very least supply an --ami name, as well as the --role and --s3-bucket that were installed in your gem's deployment.
    * Supplying a --key-pair-name is recommended for debugging over RDP.
    * --subnet-id and --security-group-id can come from your gem's deployment, to prevent auto-creation every time you run this script.
    
CREATE A SHUFFLED DICTIONARY

1. From the Harness directory, run shuffle_words.py to create shuffled_words.zip from words.zip.

USE THE CGP TO UPLOAD YOUR DATA AND CONFIGURE YOUR WORKFLOW

1. Log into the CGP for your project and deployment, and view the Cloud Gem Compute Farm.
2. Go to the Build Configuration tab.
3. In the s3_file field, enter the name of the zip file you want to process without the ".zip", i.e. "shuffled_words".
4. [OPTIONAL] In the s3_dir field, enter a directory name to use, e.g. "foo".
5. In the max_level field, enter the maximum depth of recursive division you want to divide the problem into. (2 or 3 is good for testing purposes.)
6. Click the "Upload Data" button.
7. Browse to and select "shuffled_words.zip".
    * If you entered a directory in the s3_dir field, you need to prepend that directory to your destination path/key, e.g. "foo/shuffled_words.zip".
8. Click "Upload" to upload the dictionary to S3.

TEST YOUR HARNESS LOCALLY

On the Overview tab you can hit Run Workflow. Then you can run a Decider and Worker locally on the command line to test them out.

1. When running from the command line, you need to supply all of the parameters from your resource group deployment.
    * --domain is the SWF domain that was generated in your resource group, i.e. {Workflow Physical Id}-domain
    * --task-list is the TaskList.name supplied for the Workflow in the resource template
    * --div-task, --merge-task, --build-task are all taken from the "ActivityTypes" of your Workflow in the resource template
    * --role is the ARN for the InstanceRole generated in the resource group
    * --log-db is the table name for the LogDB generated in the resource group
    * --config-bucket is the name of the computefarm S3 bucket in the resource group
2. You must one run instance of a decider (-rd or --run-decider), and then one or more instances of a worker (default behavior).

TEST YOUR HARNESS WITH A FLEET OF EC2 INSTANCES

1. Go to the Fleet Management tab of the CGP.
2. Click on "New Fleet".
3. Give your fleet a unique configuration name.
4. Choose the AMI you created earlier.
5. [OPTIONAL] If you specified a --key-pair-name when building your AMI, choose it here so you can log in to your instances via RDP.
6. Choose an instance type to use for this fleet.
7. Click "Continue" and proceed to the next dialog.
8. Give your fleet a unique group name (a variant of your configuration name is recommended).
9. Specify a number of instances to launch immediately (you may enter 0).
10. Choose whether you want your fleet to automatically terminate (i.e. go to 0 instances) when a workflow is completed.
11. Click "Create" to create your fleet.
12. If you did not specify any instances to immediately launch, enter a positive number now and choose "Save Changes".
13. Return to the Overview tab and choose "Run Workflow".
14. As your instances gradually come online (it may take 5-10 minutes) you will see progress occurring.

DOWNLOAD THE RESULTS

1. Once the workflow is completed, go back to the Build Configuration tab.
2. Click on "Download Build Result".
3. Locate the file "shuffled_words_sorted.zip" and download it; observe the dictionary has been sorted.
4. You may also download any of the intermediately-generated files.

CUSTOMIZE THE HARNESS TO YOUR APPLICATION

1. You may revise or replace any of divide.py, build.py and merge.py to solve other distributed problems.
2. You may alter AWS/cgp-resource-code/src/workflow-config.ts to add different parameters for configuration in the CGP used by your script.
3. Use the manifest.json to customize what software is installed on your AMI and run any PowerShell commands that are required to configure it.