loci-excelerator

A reapportionment tool for the Loc-I project

IderDown

A tool for downloading subsets of ID's based on within relationship in a given dataset. This feature can be found at the /iderdown url. ie Get me all meshblocks within a given SA2

Development

Repository Structure

/excelerator - Contains a meteor JS (http://meteor.com) application /.deploy - Contains deployment files using meteor up and docker as the main deployment platform

Excelerator Meteor Application

This is standard meteor app using Blaze.js as the rendering engine.

The basics of the app are to:

• have a user upload a CSV file,
• pick some run parameters,
• submit a job to the job queue
• show progress of the jobs
• send a notification to the user when complete
• allow the output to be downloaded

File Upload

Files are uploaded to the directory specified in the settings.json file

Job Manager

We have a simple job managed with a single worker integrated into the meteor server. It is possible to have multipl workers that take and run jobs if required. See https://github.com/vsivsi/meteor-job-collection

REST API

Incomplete It would be nice to have this service accessible via a restful API to integrate into other applications.

Deployment (dev/internal)

Deployment is managed via Meteor Up. An application to bundle and configure meteor application on remote servers.

The deployment are triggered from push to certain branches in the repository. Simply merge in master to the deployment branch of your choice and within 15 minutes the jenkins server (https://bigbrother.it.csiro.au:9443) should have worked out there are changes and started deployment.

Local development/deployment via Meteor

Install meteorJs via https://www.meteor.com/developers/install

Edit the public.uploads.path property in the settings.json file to point to a valid local file directory.

$ cd loci-excelerator
$ cd excelerator
$ meteor npm install
$ meteor --settings settings.json

Excelerator can now be accessed via http://localhost:3000

Dev / Staging

At this stage we only have prod deployment stage.

Production

Deployment is initiated via a commit to the prod_deploy branch.

The ./deploy.sh script is run to call meteor up. The keys to the EC2 instance are added as a secretFile in the jenkins instance and retrieved via the script as an env variable. The deployment is managed via docker with the following volumes on the host:

• /opt/mongo The storage for the database data
• /opt/data/uploads The location for the uploaded files from the meteor application

Deployment (Docker)

We also provide a docker container that is capable of running the application.

One of the main considerations is that uploaded files will be put in the /files directory in the container. It is currently mapped to named volume (stored on the host machine) so the container itself doesn't grow too big, and the files will be persisted between deployments. You may wish to remap this volume to somewhere else.

To stand up an instance of the application, simply run docker-compose up from the /docker directory.

It will spin up two containers, mongo, and a nodejs container running the web app.

Use docker hub images

To stand up an instance of the application using a pre-built image, run:

$ cd docker/
$ docker-compose -f docker-compose -f docker-compose.useimages.yml up -d

Pointing to other instances of the Loc-I Cache and Loc-I integrated api

To point Excelerator to different instances of the Loc-I Cache and the Loc-I integrated API, simply edit the settings.json file in the graphdb and integrationApi sections of that file.

Once the file is edited, redeploy using docker.

Testing

The methods for writting test are described at https://guide.meteor.com/testing.html#unit-testing Also see https://atmospherejs.com/meteortesting/mocha for more information on variable used to teh run the scripts

The basic commands for running the tests that I have setup so far are:

TEST_WATCH=1 MOCHA_GREP=IDerDown meteor test --settings=settings.json --driver-package meteortesting:mocha