Deployment

Corset is distributed as a set of Docker containers which can be managed using docker-compose.

docker-compose up -d

Docker containers

The table below shows relevant information about each Docker container.

Container name	Container image	Root	Port	Data path	Depends on
dp-solr	solr:latest		8983	/var/solr
dp-postgres	postgres:latest		5432	/var/lib/postgresql/data
dp-back	dp-back:latest	/opt/dp/back	5000	/opt/dp/back/data	dp-postgres
dp-front	dp-front:latest	/opt/dp/front	5000	/opt/dp/front/data	dp-postgres dp-back

The only port that needs to be public is dp-front:5000, which is the entrypoint for the web app.

This web app stores relevant information in each data path. These paths should be mounted as docker volumes for their contents to persist.

For dp-back, this is the data folder structure:
- data/logs: API gunicorn logs, Celery logs
- data/redis-data: data used by redis
- data/corsets: generated custom corpora
For dp-front, this is the data folder structure:
- data/logs: Gunicorn logs

Some settings for dp-back and dp-front are defined in back/app/config.py and front/app/config.py, respectively.

dp-back settings (back/app/config.py):
- CORSET_SIZE: object containing an amount of sentences for each corset size: small, medium and large.
- CORPORA_LINE_LIMIT: maximum amount of sentences to process from user uploaded files (query corpus/sample text)
- CORPORA_SAMPLE_SIZE: maximum amount of sentences that will be extracted from a corset to generate an excerpt for preview
dp-front settings (front/app/config.py):
- ENABLE_NEW_LOGINS: if true, lets new users log in. If false, only users that already logged in in the past will be able to sign in.
- BANNED_USERS: list of users that should not be able to log in.
- ADMINS: list of users with admin privileges
- GOOGLE_OAUTH_CLIENT_ID: Client ID for Google Login
- GOOGLE_OAUTH_CLIENT_SECRET: Client Secret for Google Login

Some settings are set through environment variables. The table below lists the required variables for each container and their possible values.

SECRET_KEY must have the same value in both dp-back and dp-front

Container name	Variable	Default value	Description
dp-solr	SOLR_JAVA_MEM	-Xms8g -Xmx8g	Memory configuration for Solr
dp-postgres	POSTGRES_DB	dpdb	Database name. The container will create this database automatically when launched
	POSTGRES_USER	dpdb	Username and password used to access the database. The container will set these credentials up automatically when launched
	POSTGRES_PASSWORD
dp-back	DATABASE_URL		URL to access the database in the form: `postgresql://[USER]:[PASSWORD]@dp-postgres:5432/dpdb`
	SOLR_URL	http://dp-solr:8983/solr	URL to access the Solr platform
	SOLR_USR		If your Solr instance is protected with HTTP Basic Auth, specify credentials here
	SOLR_PWD
	DEBUG	0	Enable debug mode (live code refresh)
	CORS_ORIGINS	http://localhost:5000	Add allowed CORS origins (separated by `\|`). The default value should be enough for a standard deployment
	SECRET_KEY	development key	Flask Secret Key, which determines legitimacy of sessions
dp-front	DATABASE_URL		URL to access the database in the form: `postgresql://[USER]:[PASSWORD]@dp-postgres:5432/dpdb`
	API_URL	http://dp-back:5000	URL to access the API
	DEBUG	0	Enable debug mode (live code refresh)
	SECRET_KEY	development key	Flask Secret Key, which determines legitimacy of sessions