Flowintel, an open-source case management platform

Deployment and configuration

Flowintel combines a Flask application, a relational database, Valkey/Redis sessions, a frontend asset build, and optional services such as MISP modules. A working web process alone does not confirm that reminders, enrichment or synchronisation are running.

Configuration files

FilePurpose
conf/config.pyApplication settings, database/session configuration, authentication and feature settings.
conf/config_module.pyConnector origin URL, notification delivery and module settings.
.env or service environmentRuntime values such as database credentials, SECRET_KEY and initial account settings.
docker/docker-compose*.ymlContainer services, networks, volumes and configuration mounts.

Copy configuration from the .default files for a fresh installation. For an existing installation, compare new defaults with your local settings instead of overwriting them.

Settings to review

  • SECRET_KEY: set a unique, persistent secret for the instance.
  • DB_HOST, DB_PORT, DB_NAME, DB_USER, DB_PASSWORD, DB_DIALECT, DB_DRIVER: select and connect to the database used by your runtime configuration.
  • VALKEY_IP, VALKEY_PORT, VALKEY_PASSWORD: connect to session storage.
  • ADMIN_MAIL, ADMIN_PASSWORD: configure the initial account before database initialisation; these are not a password-reset mechanism for existing users.
  • FLOWINTEL_APP_HOST, FLOWINTEL_APP_PORT: select the application bind address and port.
  • MISP_MODULE_HOST, MISP_MODULE_PORT: locate the enrichment service.
  • ORIGIN_URL in module configuration: set the origin used in exported links.
  • MISP_EXPORT_FILES: explicitly decide whether MISP exports include attachments.

The launcher selects development or production mode. -l is development; -p runs Gunicorn with production configuration. A Docker deployment selects its environment through Compose and the Docker launch path.

Native production installation

Prepare the database, runtime credentials and service account before initialisation. With configuration in place, the safe installer supports:

bash install.safe.sh --production

Production initialisation loads reference data without the development sample cases. Build frontend assets as described in Getting started, then configure process supervision, HTTPS and a reverse proxy for shared access.

The application repository provides an installation manual, NGINX example and systemd service examples under doc/. Adapt their paths, environment and service users to your host. Do not run duplicate launcher workers and independently supervised workers for the same job.

Container variants

The published-image Compose file is docker/docker-compose.yml. Local-image variants are supplied for PostgreSQL and MariaDB:

Variant in filenameServices supplied
fullFlowintel, database and Valkey
infraDatabase and Valkey for an application running on the host
nodbFlowintel and Valkey, with an external database
singleFlowintel only, with external dependencies

Local variants expect a locally built image. Run make help in the application repository to inspect build and deployment targets. For example, make build_latest_local builds the local image; use the matching Compose variant and environment template for your database.

Treat the supplied stacks as starting configurations. Review image versions, network ranges, port bindings, secrets and storage before deploying. latest is mutable; record the deployed image identity so upgrades can be reproduced.

Persistent data

Preserve the database and the files required to interpret it: uploads, configuration, history, logs where required, and any signing/encryption keys. Compose uses named volumes for several of these locations and bind mounts for configuration.

Stopping containers with docker compose down normally retains named volumes. Adding --volumes, or invoking volume-removal Make targets, deletes persistent storage. Do not use those operations as an upgrade procedure.

Continue with Operations and backups before accepting production data.