If you want to do system tests in your service and need the datastore, use the dc.external.yml. It is completely independent from the local code. Just merge it with your service's docker compose file and you can test everything in conjunction. It uses the productive setup, so no hot reload or similar is used, but runs the datastore in dev mode, so dev utils like truncate_db (see below) are available. You have to add the reader/writer as a dependency to your service(s) and add them to the datastore network.
You can issue commands to the datastore on startup via the docker variable COMMAND (has to be given as a build argument to the docker file). Currently only commands for the writer are supported.
create_initial_data: this will load the file given in the environment variableDATASTORE_INITIAL_DATA_FILE, which may be a relative or absolute path to a file inside the docker container or an URL to a web resource (e.g. the OpenSlides 4 example data) and write it to the datastore. If any data is present already, the command will fail. Note that if a path is given, the respective file has to be mounted to this place in the container; if an URL is given, the internet has to be accessible from inside the docker container (which means that in a setup with isolated networks e.g. as in the main repository, this option will not work).
Another possibility is to provide initial-data: as prefix followed by JSON encoded data as value of that environment variable if you do not want to provide a web ressource or mount a file into the container.
The script is also available from inside the container under cli/create_initial_data.py. Example setup to load initial data:
- set
COMMAND=create_initial_dataandDATASTORE_INITIAL_DATA_FILE=https://raw.githubusercontent.com/OpenSlides/openslides-backend/main/global/data/example-data.jsoninside the docker container, e.g. in yourdocker-composefile.
This will not work if the database is not empty. In this case, you can make use of the truncate_db route (see below). Example:
make run-dev
curl --header "Content-Type: application/json" -d '' http://localhost:9011/internal/datastore/writer/truncate_db
docker compose -f dc.dev.yml exec writer ./entrypoint.sh bash
export DATASTORE_INITIAL_DATA_FILE=https://raw.githubusercontent.com/OpenSlides/openslides-backend/main/global/data/example-data.json
python cli/create_initial_data.py
exit
If the variable OPENSLIDES_DEVELOPMENT is set inside the docker container, the datastore runs in development mode. This is the case by default if you use the dc.dev.yml setup (see the environment section for the writer). Again, this currently has only an influence on the writer.
In development mode, all passwords (e.g. for the database) default to openslides.
In development mode, the route truncate_db is available in the writer. This will truncate the entire database and erase all data without confirmation, so use with caution. Example curl call:
curl --header "Content-Type: application/json" -d '' http://localhost:9011/internal/datastore/writer/truncate_db
Since the folder structure inside the docker container differs from the real one, IDEs like VS Code can't follow the imports correctly. To solve that, if you use VS Code, you need to create a .env file preferably in the .vscode folder (adjust your settings variable python.envFile accordingly) with the following entry:
PYTHONPATH=shared:reader:writer
Since docker compose uses the .env file in the root of the repository, this file should not also be used by VS Code, so it has to be placed elsewhere.
For other IDEs there are probably similar solutions. Feel free to add them here for your IDE.
See Locking.