# DigitalDownConverter — Deployment

## Debian Packages

The pipeline is distributed as Debian packages built by CI and hosted in the MPIFR Aptly
repository. The `edd_dbbc` Docker image installs them at build time.

| Package | Contents |
|---|---|
| `python3-edd-dbbc` | Python library (`edd.dbbc`) |
| `edd-dbbc-utils` | Compiled C++ binaries (`ddc_processing`, etc.) |

## Docker Image

| Property | Value |
|---|---|
| Image name | `edd_dbbc` |
| Base image | `edd_build:cuda12.6.1_ubuntu24.04` |
| Ansible build role | `roles/edd_dbbc/` |
| Registry | `gitlab-registry.mpcdf.mpg.de/mpifr-bdg/edd/dbbc` |

The image installs `python3-edd-core`, `python3-edd-dbbc`, `edd-dbbc-utils`, `mkrecv`,
`mksend`, and `psrdadacpp` from the MPIFR Aptly repository.

## CI Pipeline

<!-- AI-GENERATED:START name="deployment_ci" -->
| Stage | Job | Description |
|---|---|---|
| `build` | `build_source` | CMake + make, produces `release/` artifacts |
| `build` | `build-deb` | Builds Debian packages via `.build_template` |
| `test` | `test_cpp` | CTest suite against built binaries |
| `test` | `test_python` | Python test suite with coverage |
| `stats` | `coverage_cpp` | lcov C++ coverage badge |
| `stats` | `coverage_python` | Python coverage badge |
| `stats` | `cpplint` | C++ style lint badge |
| `stats` | `pylint` | Python lint |
| `deploy` | `upload-deb` | Uploads packages to Aptly (requires build + tests) |
| `deploy` | `build_docker` | Builds and pushes `edd_dbbc` image |
<!-- AI-GENERATED:END -->

## Ansible Deployment

The pipeline is deployed via the `ddc_processor` Ansible role:

```bash
ansible-playbook provision.yml --tags ddc_processor
```

The role delegates to `EDD.core.common` with:

```yaml
image_name: edd_dbbc
default_container_name: ddc_processor
container_cmd: "edd_ddc_processor --host={{ edd_subnet }} --port={{ bind_port }}"
register_as_pipeline: True
provision_test_imports: "edd.dbbc.ddc_processor_tests"
```

## Provision Tests

Provision tests are imported from `edd.dbbc.ddc_processor_tests`. They verify that the
pipeline starts, accepts configuration, and reaches the `configured` state before the
observation begins.