Wave Apps on Heroku

Wave makes it very easy to create data science applications. Heroku is a cloud platform where we can quickly deploy our Wave app and share it with the rest of the world. To know more about Heroku, read What is Heroku ?

Following are the step by step instructions to deploy your Wave app to Heroku platform. A GitHub repository with an example Wave app that can be deployed to Heroku is available here.

And yes, we can deploy our Wave apps for free on Heroku.

Table of Contents

• Requirements
• Overview
• Step by Step
• 1. Create a Wave App
  • Clone the repo
  • Setup Virtual Environment
  • Run the Wave App
• 2. Dockerize the Wave app
  • Get Dockerfile
  • Build the Docker Image
  • Run the Wave app using Docker
• 3. Deploy the Docker image to Heroku
  • Log into Heroku using CLI
  • Log into Container Registry
  • Create a Heroku app
  • Tag the Docker Image
  • Push the image to Heroku
  • Release the image to your app
  • Open the app in your browser
• What is Next?
• Feedback

Requirements

Before we begin, please make sure you have these requirements covered.

• Docker
• Heroku Account
• Heroku CLI

Overview

1. Create a Wave app and make sure it is running on our local machine
2. Dockerize the Wave app
3. Deploy the Docker image to Heroku
4. Cake

Step by Step

Let's begin.

1. Create a Wave App

You can create your own Wave app (examples) or use an existing one. For the sake of this tutorial, let's clone from srini-x/wave-heroku

If you already have an app ready to go, you can skip to the next section.

Clone the repo

$ git clone https://github.com/srini-x/wave-heroku.git
$ cd wave-heroku

Setup Virtual Environment

$ python -V
Python 3.8.11

$ python -m venv .venv
$ . .venv/bin/activate
(.venv)$ pip install --upgrade pip
(.venv)$ pip install -r requirements.txt

Run the Wave App

Let's run our app on our local machine. If you are using Wave version older than 0.20.0, you need to start the Wave server (waved) before running the Python app. Please refer to the Wave Docs.

(.venv)$ wave run ai_app/app.py

2022/02/04 19:13:52 #
2022/02/04 19:13:52 # ┌────────────────┐ H2O Wave
2022/02/04 19:13:52 # │  ┐┌┐┐┌─┐┌ ┌┌─┐ │ 0.20.0 20220131055911
2022/02/04 19:13:52 # │  └┘└┘└─└└─┘└── │ © 2021 H2O.ai, Inc.
2022/02/04 19:13:52 # └────────────────┘
2022/02/04 19:13:52 #

app will be available at http://localhost:10101

2. Dockerize the Wave app

Get Dockerfile

For this step we need to add 3 new files to our repo. All three files are available at srini-x/wave-heroku. Place all three files at the root of the repo/project.

• Dockerfile - This is the file we need to build our Docker image.
• docker-entrypoint.sh - This is the bash script that is called to start out Wave app inside the Docker container.
• .dockerignore - This file is similar to .gitignore file which contains the list of all unnecessary files.

If you are following along from the top and cloned the example repo, the files are already there.

Note

Please make sure there is a requirements.txt file at the root of the repo/project. If you don't have one, please generate one using the following command

While the virtual environment is activated, run

(.venv)$ pip freeze > requirements.txt

Build the Docker Image

Be at the root of the repo and run this command to create a Docker image with our Wave app inside it. Change the values of PYTHON_VERSION, WAVE_VERSION, and PYTHON_MODULE to match your setup. PYTHON_MODULE is same as the last part of the wave run <PYTHON_MODULE> command.

$ docker build \
--build-arg PYTHON_VERSION=3.8 \
--build-arg WAVE_VERSION=0.20.0 \
--build-arg PYTHON_MODULE="ai_app/app.py" \
-t wave-heroku:0.1.0 .

Check that our image is available

$ docker image ls

REPOSITORY                                 TAG                        IMAGE ID       CREATED         SIZE
wave-heroku                                0.1.0                      84621557615e   2 minutes ago     457MB

Run the Wave app using Docker

Let's start a Docker container from the Docker image we created.

Note:

In the following command, you can pick any PORT values for -p and -e PORT. This -e PORT=<waved-port-inside-container> option is used by Heroku to set the PORT for our waved server. When running locally, we can select any value for PORT as long as it is available. However, we need to use the same PORT value for -p option. Ex: -p <app-port-outside-container>:<waved-port-inside-container>

If we select our PORT inside the container to be 9999 and port outside the container to be 8080, the command will look like -p 8080:9999 -e PORT=9999. In this case, the app will be available at http://localhost:8080.

You can also select same PORT for inside and outside, Ex: -p 10101:10101 -e PORT=10101.

$ docker run --rm --name wave-heroku -p 10101:8888 -e PORT=8888 wave-heroku:0.1.0

$ ( cd /home/appuser/wave/wave-0.20.0-linux-amd64 && ./waved -listen ":8888" & )

2022/02/05 03:34:26 #
2022/02/05 03:34:26 # ┌────────────────┐ H2O Wave
2022/02/05 03:34:26 # │  ┐┌┐┐┌─┐┌ ┌┌─┐ │ 0.20.0 20220131055911
2022/02/05 03:34:26 # │  └┘└┘└─└└─┘└── │ © 2021 H2O.ai, Inc.
2022/02/05 03:34:26 # └────────────────┘
2022/02/05 03:34:26 #
2022/02/05 03:34:26 # {"address":":8888","base-url":"/","t":"listen","web-dir":"/home/appuser/wave/wave-0.20.0-linux-amd64/www"}

$ wave run --no-reload --no-autostart ai_app/app.py

2022/02/05 03:34:30 #
2022/02/05 03:34:30 # ┌────────────────┐ H2O Wave
2022/02/05 03:34:30 # │  ┐┌┐┐┌─┐┌ ┌┌─┐ │ 0.20.0 20220131055911
2022/02/05 03:34:30 # │  └┘└┘└─└└─┘└── │ © 2021 H2O.ai, Inc.
2022/02/05 03:34:30 # └────────────────┘
2022/02/05 03:34:30 #
2022/02/05 03:34:30 # {"address":":10101","base-url":"/","t":"listen","web-dir":"/home/appuser/venv/www"}
INFO:     Started server process [1]
INFO:     Waiting for application startup.
2022/02/05 03:34:31 # {"host":"http://127.0.0.1:8000","route":"/","t":"app_add"}
INFO:     Application startup complete.
INFO:     Uvicorn running on http://127.0.0.1:8000 (Press CTRL+C to quit)

app will be available at http://localhost:10101

3. Deploy the Docker image to Heroku

Ok. Hard part is over. Let's login to Heroku and deploy our app.

Log into Heroku using CLI

Once we have Heroku CLI installed, we can login to Heroku using the following command. This will open a browser and ask you to login to Heroku

$ heroku login

heroku: Press any key to open up the browser to login or q to exit:
Opening browser to https://cli-auth.heroku.com/auth/cli/browser/77d5068e-0r48-34g2-uiw9-0805c7280efe?requestor=SFMyNTY.g2hoqGFFFF02Ny4xODAuMTA1Ljc2bgYACBy8x34BYgABUYA.uF3ICR3g2gehvgFiblrd6jfHRIiHbAXZkGbmUFxXICg
Logging in... done
Logged in as <your-email@mail.com>

Log into Container Registry

$ heroku container:login

Create a Heroku app

Be in the root of the repo and run

$ heroku create

Creating app... done, ⬢ hidden-reef-76107
https://hidden-reef-76107.herokuapp.com/ | https://git.heroku.com/hidden-reef-76107.git

Optionally, rename the app to give it a friendly name. Make sure that the name is unique

$ heroku apps:rename --app hidden-reef-76107 my-wave-app-1234

Renaming hidden-reef-76107 to my-wave-app-1234... done
https://my-wave-app-1234.herokuapp.com/ | https://git.heroku.com/my-wave-app-1234.git
 ▸    Don't forget to update git remotes for all other local checkouts of the app.
Git remote heroku updated

Tag the Docker Image

Get the <app-name> from the previous step. You can also run heroku apps command to get the list of app names. In this example, our Heroku app name is my-wave-app-1234.

Use the following command to give a new tag to our existing Docker image. Please replace <app-name> below.

$ docker tag wave-heroku:0.1.0 registry.heroku.com/<app-name>/web

Push the image to Heroku

Push our docker image to Heroku's registry.

$ docker push registry.heroku.com/<app-name>/web

Release the image to your app

Finally, tell Heroku to deploy your app and make it publicly available on the internet.

$ heroku container:release web

Open the app in your browser

Open your Wave app running on Heroku platform. The following command will automatically open the app in your default browser.

$ heroku open

To get the Web URL of the app, run

$ heroku apps:info

=== my-wave-app-1234
Auto Cert Mgmt: false
Dynos:          web: 1
Git URL:        https://git.heroku.com/my-wave-app-1234.git
Owner:          <your-email@mail.com>
Region:         us
Repo Size:      0 B
Slug Size:      0 B
Stack:          container
Web URL:        https://my-wave-app-1234.herokuapp.com/

What is Next?

• Learn more about Wave from the Docs
• Learn more about deploying on Heroku at Deploying with Docker
• Also, checkout the blog post about deploying Wave apps on AWS Lightsail or EC2

Feedback

We are eager to hear your feature requests, bugfixes or just general feedback in our discussions and issues sections.