Skip to content

Latest commit

 

History

History
316 lines (214 loc) · 12.7 KB

DOCKER.md

File metadata and controls

316 lines (214 loc) · 12.7 KB
Raw

ANYWAY’s docker environment

Docker is an open source project to pack, ship and run any application as a lightweight container.

The idea is to deploy a container (light weight environment) that has all our app dependencies installed and ready to go. As a developer you can do one of both:

  • Use the container do write your code (using vim/nano or any other GNU text editing tool - not recommended)
  • Use a git repo cloned to your local machine and use the container as a remote server running the app

The container loads itself with the command given in the instructions, it has the DB on it, deployed and ready to work (at /anyway/local.db) With every local code change, the container would restart itself and the changes would be immediately available in your web browser. The current image is based on an Ubuntu linux with java.

Read more on the docker Github project

Instructions

NOTE: If you're using windows, complete the "getting the code" section after installing WSL, inside the new ubuntu environment.

1. Get the code

2. Install Docker and Install Docker Compose

3. Get the .env file with the required secret values and place in the project root directory - can be downloaded from here. Note that this file needs to be saved as .env - with the . at the beginning of the name of the file. Continue with your OS, See below

For Mac:

4. If this is your first time installing ANYWAY Docker environment - move on to stage 5. Otherwise, to build an existing environment with the most updated DB, remove DB volume by running docker volume rm anyway_db_data. Note - this will delete all of your local DB data!

5. Anyway images stored on github package, to be able to pull the images you need to login to github using docker login. if you already logged in with docker to github source jump to the next step.

docker login

First, to login to github go to this link of your github account and generate token with read:packages permission.

Second, copy the token that you just generate and run docker login command like this when USERNAME is your github username, and paste the token when prompt will ask the password.

$ docker login docker.pkg.github.com -u USERNAME

6. Go to the project's root directory and run: sudo docker-compose up anyway This will start the containers. It will take a few minutes until it's done.

7. You're all set! ANYWAY is up and running with the DB data - connect to http://127.0.0.1:8080 Note - you won't see the map since the key works in production. If you need to see the map contact atalya via slack to get a developer key.
The developer key need to replace the production key in the file /anyway/blob/dev/templates/index.html where you can find: "https://maps.googleapis.com/maps/api/js?key=AIzaSyDUIWsBLkvIUwzLHMHos9qFebyJ63hEG2M&libraries=places,visualization&language=iw" (google maps url) So if the developer key is "12345" the new url need to be is: "https://maps.googleapis.com/maps/api/js?key=12345&libraries=places,visualization&language=iw"

8. To stop the containers run: docker-compose down

9. To restore fresh DB data, delete all existing volumes: docker-compose down -v then restart from step 6

For Windows: You have three options: Use WSL and install under WSL. Use WSL and install under windows, or install ubuntu VM (See UBUNTU_VM_ON_WINDOWS). For WSL - see also ubuntu for wsl

Important: Git uses CRLF line ending under Windows, by default. Make sure shell and dockerfiles use LF line ending. This means at least the following files:

  • Dockerfile
  • docker-entrypoint.sh
  • db_docker/Dockerfile
  • db_docker/dumpdb.sh
  • db_docker/restore_db.sh
  • bin/docker_login.sh
  • bin/nginx_docker_build_push.sh Alternatively, in order to evade line ending issues, clone the repository inside WSL terminal.

Pick one option and continue to the instructions section For Ubuntu.

For Ubuntu:

4. If this is your first time installing ANYWAY Docker environment - move on to stage 5. Otherwise, to build an existing environment with the most updated DB, remove DB volume by running sudo docker volume rm anyway_db_data. Note - this will delete all of your local DB data!

5. Anyway images stored on github package, to be able to pull the images you need to login to github using docker login. if you already logged in with docker to github source jump to the next step.

docker login

First, to login to github go to this link of your github account and generate token with read:packages permission.

Second, copy the token that you just generate and run docker login command like this when USERNAME is your github username, and paste the token when prompt will ask the password.

$ sudo docker login docker.pkg.github.com -u USERNAME

6. Go to the project's root directory and run: sudo docker-compose up anyway This will start the containers. It will take a few minutes until it's done.

7. You're all set! ANYWAY is up and running with the DB data - connect to http://127.0.0.1:8080 Note - you won't see the map since the key works in production. If you need to see the map for development email us [email protected] to get a developer key.
The developer key need to replace the production key in the file /anyway/blob/dev/templates/index.html where you can find: "https://maps.googleapis.com/maps/api/js?key=AIzaSyDUIWsBLkvIUwzLHMHos9qFebyJ63hEG2M&libraries=places,visualization&language=iw" (google maps url) So if the developer key is "12345" the new url need to be is: "https://maps.googleapis.com/maps/api/js?key=12345&libraries=places,visualization&language=iw"

7. To stop the containers run: sudo docker-compose down

8. To restore fresh DB data, delete all existing volumes: docker-compose down -v then restart from step 7

Using VSCODE

In order to use VSCODE in debugging mode with DOCKER, check out VSCODE_CONFIGURATION

Working on anyway-etl and Airflow

Anyway ETL processes and Airflow server are developed in a different repository: hasadna/anyway-etl but we use the same Docker Compose environment.

For some tasks you will also need to set secret values in the .env file, ask a team member for these values.

Running anyway-etl tasks

  • Pull the latest anyway-etl Docker image: docker-compose pull anyway-etl
  • See the available anyway-etl commands: docker-compose run anyway-etl --help
  • Run a command: docker-compose run anyway-etl cbs parse-all

Developing anyway-etl tasks

To develop anyway-etl using the Docker Compose environment you first need to clone hasadna/anyway-etl. The clone should be a sibling directory to anyway, so it will be at ../anyway-etl relative to anyway repository.

  • Build the anyway-etl Docker image: docker-compose -f docker-compose.yml -f ../anyway-etl/docker-compose-override.yaml build anyway-etl
  • Run anyway-etl commands: docker-compose -f docker-compose.yml -f ../anyway-etl/docker-compose-override.yaml run anyway-etl --help
    • When running this command, any changes you make to anyway or anyway-etl code will take effect immediately without requireing to rebuild the image.

Running the Airflow server

  • Pull the airflow server image: docker-compose pull airflow-webserver airflow-scheduler
  • Start the server: docker-compose up -d airflow-scheduler
  • Access the server at http://localhost:8082 with username admin and password 123456

Developing the Airflow server

  • Build and start the server: docker-compose -f docker-compose.yml -f ../anyway-etl/docker-compose-override.yaml up -d --build airflow-webserver airflow-scheduler
    • On first run it may take a while for the server to be available

Running the ETL Nginx server

The ETL Nginx servers allows to browse the data used by the ETL tasks

  • Pull the image: docker-compose pull anyway-etl-nginx
  • Run the server: docker-compose up -d anyway-etl-nginx
  • Browse the files at http://localhost:8083

Additional Docker commands

Use sudo before each docker commands if you are using ubuntu.

To install requirements again, update DB schema OR redeploy any of the requirement involving the dependencies installation, simply rebuild the image - Run in anyway directory:

docker-compose build

Run docker-compose in detached mode - containers are running in background:

docker-compose up -d anyway

List your local docker images:

docker images

List your running containers:

docker ps

List all containers ever:

docker ps -a

Stop a running container (id is listed in docker ps):

docker top <container-id>

Rebuild the image:

docker-compose build

Open container bash terminal to execute commands (while container is running):

docker exec -it <container-name> bash

For example - to open the bash terminal of the db container:

docker exec -it db bash

For example - to open the bash terminal of the anyway container:

docker exec -it anyway bash

Inside the bash terminal of anyway container - run the following to use postgres command line:

psql $DATABASE_URL

Be careful with the following command Deleting an image(from docker images):

docker rmi <image-id OR image repository>

Be careful with the following command Remove the DB volume (this will delete all db data)

docker volume rm anyway_db_data

Be careful with the following command Delete all docker unused images and all volumes

docker system prune --all --volumes

Set your VM with the current running shell session:

eval "$(docker-machine env default)"

Additional Notes

If you want to work with your own code and the docker OR if your 8080 port is already taken by other dockers\servers you need to create an override for the docker-compose. For example:

services:
 anyway:
   volumes:
     - ./anyway:/anyway/anyway

This loads the ./anyway dir (relative to the docker-compose file) as /anyway/anyway in the docker overriding the inner volume and allowing you to run your own code inside the docker.

Questions and ideas

Talk to Atalya on HASADNA's Slack (atalya) or email us [email protected].

Testing production environment locally

This process allows to emulate a full production environment locally for testing. This is an advanced operation and not needed for normal development.

Create a .env file for production (set relevant values):

# app env vars
DATABASE_URL=postgresql://anyway:12345678@db/anyway
GOOGLE_MAPS_KEY=
TWITTER_CONSUMER_KEY=
TWITTER_CONSUMER_SECRET=
TWITTER_ACCESS_KEY=
TWITTER_ACCESS_SECRET=
FACEBOOK_KEY=
FACEBOOK_SECRET=
GOOGLE_LOGIN_CLIENT_ID=
GOOGLE_LOGIN_CLIENT_SECRET=
MAILUSER=
MAILPASS=
GOOGLE_APPLICATION_CREDENTIALS=/secrets/GOOGLE_APPLICATION_CREDENTIALS_KEY.json
GOOGLE_APPLICATION_CREDENTIALS_HOST_PATH=/host/path/to/google_application_credentials.json
APP_SECRET_KEY=

# db env vars
POSTGRES_USER=postgres
POSTGRES_PASSWORD=123456
POSTGRES_DB=postgres
#   aws access/secret with permissions to read from full db dumps bucket
DBRESTORE_AWS_ACCESS_KEY_ID=
DBRESTORE_AWS_SECRET_ACCESS_KEY=
DBRESTORE_AWS_BUCKET=dfc-anyway-full-db-dumps
DBRESTORE_FILE_NAME=2020-06-09_anyway.pgdump
#   should match the password set in app env vars
DBRESTORE_SET_ANYWAY_PASSWORD=12345678

# db-backup env vars
DBDUMP_S3_FILE_PREFIX=testing_
#   aws access/secret with permissions to write to both full and partial db dumps buckets
DBDUMP_AWS_ACCESS_KEY_ID=
DBDUMP_AWS_SECRET_ACCESS_KEY=
#   db connection details to the postgres user
DBDUMP_USER=postgres
DBDUMP_PASSWORD=123456
DBDUMP_HOST=db

Create a shell alias to run docker-compose for production

alias docker-compose-prod="docker-compose -f docker-compose.yml -f docker-compose-production.override.yml"

Restore the DB

docker-compose-prod down -v
docker-compose-prod up --build db

Start the app

docker-compose-prod up --build nginx anyway

Access the app at http://localhost:8000

Run the backup job

docker-compose-prod build db-backup
docker-compose-prod run db-backup