Setting Up DroneTM For Development¶

Info

This guide is for local development. For production / self-hosted deployments, see the Installation Guide.

Prerequisites¶

Important

You must have Docker and Docker Compose installed first:

Just is required too:

Just

(Optional) Configure Google OAuth¶

Create a new client ID and secret in your Google account for OAuth2.
Set the redirect URL to your local frontend URL, e.g. http://localhost:3040/auth.

Note

This step is only required to use the frontend.

Configure Your Dotenv¶

There is an example .env.example file that .env can be generated from using this command: just config generate-dotenv
- The generator runs the repo’s ./envsubst twice so defaults that reference other vars (e.g. $BACKEND_WEB_APP_PORT) resolve correctly.
If you only plan on using the backend then everything should be configured for you.
The current setup expects:
- DOMAIN for public URLs (backend derives its public base URL automatically)
- VITE_API_URL for the frontend to reach the backend API (must include /api)
S3 endpoints:
- S3_ENDPOINT_UPLOAD: used by browser presigned uploads (often http://localhost:9000 in dev)
- S3_ENDPOINT_DOWNLOAD: used by browser downloads/display
  - In dev, defaults to http://localhost:9000 (downloads via a path-rewriting proxy will break presigned signatures).

If you set up Google OAuth credentials, set the variables here:

GOOGLE_CLIENT_ID="YOUR_CLIENT_ID"
GOOGLE_CLIENT_SECRET="YOUR_CLIENT_SECRET"
# Redirect URI must match your Google OAuth app config
GOOGLE_LOGIN_REDIRECT_URI="http://localhost:3040/auth"

Monitoring¶

At present, only Sentry is configured as the backend for logging with OpenTelemetry. By default logging in non-production environments is disabled.
To set it up and work with it, you will need these two ENV variables set correctly:
```
MONITORING="sentry"
SENTRY_DSN="<sentry dsn url here>"
```
Then make sure to rebuild your backend Docker image: docker compose build backend so that Docker now knows to install the extra set of dependencies. You should see a success message if it worked correctly.

LOG_LEVEL is set to info by default. Setting it to debug makes the logs very verbose, so it is recommended to leave the setting as is.

Build and Run The Containers¶

The Justfile in the project root contains all the logic needed to build and run the application.
compose.sub.yaml is used for production deploys (just start prod) - compose.yaml, meanwhile, is used for development.
To start the application, run: just start all
This command will pull the required Docker images and start all services correctly.

Access The Services¶

DroneTM Backend: http://localhost:8000

DroneTM Frontend: http://localhost:3040

Note the ports may be different if you changed them in the dotenv file.

End-to-end imagery processing (ScaleODM)¶

DroneTM submits processing jobs to ScaleODM, which reads imagery directly from S3 and writes outputs back to S3. The compose stack ships a small NodeODM container as a placeholder so the backend boots cleanly, but NodeODM cannot complete an end-to-end run: it has no S3 writer, so the finalize step (which expects the orthophoto to already be in S3) will fail. To exercise the full pipeline locally you need ScaleODM running.

Start ScaleODM on your host (kind/k3d, see the ScaleODM quickstart). It exposes its API on localhost:31100 by default.

Point the backend at it from inside the compose network. Set in .env:

# Backend container reaches the host via host.docker.internal
ODM_ENDPOINT=http://host.docker.internal:31100
# Only set this if your S3 hostname differs from what ScaleODM workflow
# pods can resolve (self-hosted MinIO/RustFS):
# SCALEODM_S3_ENDPOINT=http://host.docker.internal:9000

Restart the backend: just start backend.

When just start all runs it will probe localhost:31100/info and warn if ScaleODM isn't reachable. The warning is informational - the rest of DroneTM (uploads, ingest, classification, UI) works without ScaleODM; only the processing path needs it.