November 11, 2018

1587 words 8 mins read

Staffjoy/suite

Staffjoy/suite

Staffjoy V1, aka “Suite” - a scheduling application for hundreds of workers

repo name Staffjoy/suite
repo link https://github.com/Staffjoy/suite
homepage https://www.staffjoy.com
language Python
size (curr.) 18100 kB
stars (curr.) 787
created 2017-02-24
license Other

Suite, aka Staffjoy V1

Build Status Moonlight contractors

Staffjoy is shutting down, so we are open-sourcing our code. This version of our V1, intended for on-demand companies and call centers, has been heavily modified so that Staffjoy customers may continue using the software. It is a full workforce management suite, including time off requests, compliance management, and clock-in. Workers can even clock in and claim shifts. If you’re managing small businesses and want only basic schedulding with text message support, please use Staffjoy V2

API Documentation is available on staffjoy-suite.readme.io.

Staffjoy Suite

Credit

The authors of the original code were @philipithomas, @andhess, and @bahador. This is a fork of the internal repository. For security purposes, the Git history has been squashed.

Licensing

If you are using this repo for commercial purposes, please review the Highcharts licensing requirements for commercial use, in addition to the LICENSE file in this repo.

Demo Videos

Staffjoy Suite is open-source workforce management software for large teams.

Running the repo for production use

Database Requirements

  • Redis
  • MySQL

Environment Variables

This table intends to explain the main requirements specified in app/config.py. This configuration file can be manually edited, but be careful to not commit secret information into the git repository. Please explore the config file for full customization info.

Name Description Example Format
ENV “prod”, “stage”, or “dev” to specify the configuration to use. When running the code, use “prod”. prod
BASE_URL URL where the code is hosted. https://suite.staffjoy.com
MANDRILL_API_KEY API Key for Mandrill for sending emails.
FROM_EMAIL Email address from which notifications will be sent in Mandrill team@staffjoy.com
RECAPTCHA_PUBLIC_KEY Public key for Recaptcha
RECAPTCHA_PRIVATE_KEY Private key for Recaptcha
REDIS_HOST Host for redis localhost
REDIS_PORT Port to connect to Redis on. Defaults to 6379. 6379
REDIS_DATABASE If using multiple databases, set the number. Defaults to 0 0
SQLALCHEMY_DATABASE_URI Connection info, including username, password, and database for MySQL mysql://root:bacon@localhost/dev
SECRET_KEY A unique, secret key for your application that is used to sign cookie data. Make sure it is the same across instances in an environment. Setting this key is critical for security. (details) anyLongSecretKeyYouMakeUpAndIsRandom
STATUS_PAGE_ID Optional page id for a Status Page integration
STATUS_PAGE_API_KEY Corresponding API key for Statuspage.io
INTERCOM_ID Optional - application ID for an Intercom app
INTERCOM_SECRET Optional - Intercom secret used for signing user hashes
INTERCOM_API_KEY Optional - Intercom API key used for sending events to the Intercom API
TWILIO_ACCOUNT_SID API Key for Twilio - note that the TWILIO_NUMBER variable in config.py needs to be updated with sending phone numbers that you own for this to work.
TWILIO_AUTH_TOKEN Twilio secret key

Required and Optional Services

Cron

The /api/v2/internal/cron/ endpoint must be triggered every 60 seconds. We open-sourced our Cron microservice that does this, but consider just using a Jenkins job with:

curl --user API_KEY: http://suite.local/api/v2/internal/cron/

(NOTE: the trailing colon after the API key is required)

Chomp

Required for calculating shifts from forecasts. View Chomp’s source code on github.com/staffjoy/chomp-decomposition

Mobius

Mobius assigns workers to shifts, subject to constraints. View Mobius’s source code on github.com/staffjoy/mobius-assignment

How we deployed

  • AWS elastic beanstalk to run the Docker containers (here’s our deploy script)
  • AWS RDS for MySQL
  • AWS Elasticache for Redis
  • Healthcheck /health endpoint to determine instance health
  • Terminate SSL at load balancer
  • Run Cloudflare upstream
  • Run at least three production containers (N + 2) for redundancy
  • Stream logs to Papertrail
  • Strict firewalling of Redis for security

Notes

  • It is suggested that you fork this repository if you are using it
  • If you do not want anybody from the web signing up for your application and provisioning new organizations, modify ALLOW_COMPANY_SIGNUPS in app/config.py
  • The Staffjoy mobile applications do not work with self-hosted applications.

Development Setup

Prequisites:

Add your email to the file user.txt, e.g. if your email is lenny@staffjoy.com:

echo "lenny@staffjoy.com" > ./user.txt

This auto-registers you on first boot, and makes you an admin.

To boot a dev server for the first time, run:

make dev

The first time it launches, it will take a long time to download the required base images. After that, required packages and images will be cached. The app will be running at suite.local

To quickly launch a dev server that has already been built, you can use this shortcut command from the project root:

vagrant up
vagrant ssh
cd /vagrant
make dev-server

Be advised that this does not update any required packages or recompile the less stylesheets, so please run make dev occasionally.

Installing new packages

Add new dependencies to the end of requirements.txt, then run make dev. The dev server is booted, the dependencies are installed, then the boot script immediately runs pip freeze to update requirements.txt with the latest version number, and nested dependencies.

Hard-resetting a dev server

If you run into issues or if you messed up your environment, you can delete its caches with vagrant destroy -f. Running make dev will take longer as the system re-fetches dependencies.

Creating a sudo account in dev

After creating a Staffjoy account in your dev environt, use the commands below to access the shell:

vagrant ssh
cd /vagrant
make shell

Now in the shell window, write and execute the following code:

print User.query.get_all() # To find users - don't do this in prod where there are lots of user :-)
u = User.query.get(<id>)
u.sudo = True
db.session.commit()

The id will correspond to the id of the user that you just created and will be visible in the URL bar

Working with the Repo

Environments

To access the current environment, use:

from flask import current_app
print current_app.config.get("ENV") # dev or prod - as defined in config.py.
# Note that stage is treated as prod except for the robots.txt

Working with the database

Dev - uses MySQL just like production. To access it, run the following commands from the project root:

vagrant ssh
<wait for login>
cd /vagrant
mysql -u root -p
<enter the password "bacon">
use dev;

The development MySQL root password is bacon.

Database Migrations Summary

  1. Make a change to the model (by modifying app/models.py)
  2. Generate a migration that spits out the changes that need to be made to the database with make db-migrate
  3. Apply those generated changes with make db-deploy

Note: Commit the migration files

Dev

While your vagrant machine is running, SSH into the instance with vagrant ssh. Then in /vagrant/, run make db-migrate to generate a migration. To apply the database change, you can either run make db-deploy, or opt for a standard make-dev.

If this doesn’t work

It’s easier to wipe a database than try and do a rollback. Log into the Dev MySQL Instance and run this command to wipe it:

drop database dev;
create database dev;

Exit MySQL and now run make db-deploy to rebuild it. You will have to create a new sudo user with the Shell Context.

Shell Context

You can run python main.py shell and interact directly with the app, users, etc.

e.g.

python manage.py shell
u = User()
u.password = 'cat'
u.password_hash # See hash
u.verify_password('cat')
u.verify_password('dog')
db.session.commit()

This is useful in dev environments for clearing out the database or going into production web servers to flag a user as “sudo”.

API

Public docs forthcoming

Tokens

Tokens are time-based tokens with a default life of 6 hours. API keys are permanent until revoked.

We authenticate using HTTP basic auth, where the username is the token and the password is blank. To do this with curl, use this command: (noting that the trailing colon after the token is important because it means “no password”)

curl -u TOKEN: https://www.staffjoy.com/api/v2/

You can always get a new token for your logged-in user at /auth/api-token.

Specs

Our API uses JSON. Response data is in the data field, resources on the object are given in the resources field, and other metadata - like limit or offset may be provided.

Formatting

This library uses the Google YAPF library to enforce PEP-8. Using it is easy - run make fmt to format your code inline correctly. Failure to do this will result in your build failing. You have been warned.

To disable YAPf around code that you do not want changed, wrap it like this:

# yapf: disable
FOO = {
    # ... some very large, complex data literal.
}

BAR = [
    # ... another large data literal.
]
# yapf: enable

Miscellaneous

Resolving database issues

If you end up with multiple heads, do:

python main.py db heads
# Using the two hashes
python main.py db merge <hash1> <hash2>

Other Features

  • Twilio IVR
  • StatusPage.io integration
  • Mobile application detection via headers

Omissions

This is not a perfect copy of our internal repo. For ease of use, sanity, and security, removed parts include:

  • Billing (we used Stripe and Paid Labs)
  • Event tracking with Intercom
  • Custom code for legacy clients
  • HSTS headers - we don’t want to unintentionally trigger this on test deployments, but you can uncomment the headers in the nginx.conf
  • Dev email limits - We limited development emails to @staffjoy.com email addresses

Known issues

  • The development environment works, but really needs a ground-up rewrite to something like Docker Compose. (It hasn’t been extensively modified since before Staffjoy was a full-time job!)
  • The tests use current_app rather than the generator create_app. This should be corrected.
  • The session management system, when reading a session from a cookie, recreates the session. This means that the list of active sessions is longer than expected (though still secure).
comments powered by Disqus