Harmony GDAL Adapter (HGA)

A Harmony backend service that transforms input images using GDAL.

HGA is published to ghcr.io GitHub's Container Registry.

HGA is invoked by harmony when the harmony server is configured by UMM-Service UMM-Collection associations in CMR, to handle an incoming request for the collection. You can see examples of requests that harmony dispatches to the harmony-gdal-adapter by examining the regression test notebook for hga.

Test with Docker

Build HGA image

bin/build-image

Creates the image ghcr.io/nasa/harmony-gdal-adapter.

Build Tests

bin/build-test

Creates the nasa/harmony-gdal-adapter-test test image.

Run Tests

bin/run-test

The run-test script mounts test-reports and coverage directories and runs the test script tests/run_tests.sh inside of the Docker test container.

Test Locally

Create isolated environment

conda create --name hga python=3.10 --channel conda-forge
conda activate hga

Install requirements

conda install gdal==3.6.4
pip install -r requirements.txt -r requirements_dev.txt

Run Tests

./tests/run_tests.sh

This script runs pytest on the ./tests directory.

Debugging

It is possible to debug this service for development by attaching a debugger that follows the debugger-adapter-protocol to a harmony stack running in a local kubernetes cluster. These instructions are for developers of this service in order to help them understand the code.

Basic steps for debugging are:

1. Add debugpy to requirements.txt file and reubild this image ./bin/build-image.

   • add debugpy to the requirements.txt file:

   • requirements.txt diff:

     +debugpy==1.6.3

2. Edit your harmony .env file to use debugpy and relaunch harmony services to enable this change. The default invocation args are for this service are python -m gdal_subsetter and you must change the default params to run through debugpy listening on all interfaces at port 5678.

   HARMONY_GDAL_ADAPTER_INVOCATION_ARGS='python -m debugpy --listen 0.0.0.0:5678 --wait-for-client -m gdal_subsetter'

3. Determine the name of your service pod in kubernetes, finding the one that is named like harmony-gdal-adapter-58b6f98b57-sv5vm with different trailing hashes.

   kubectl get pods -n harmony

4. Open a port from your local machine to the kubernetes pod, subsituting your pod's name. This allows your local debugger to attach to the running process on the pod.

   kubectl port-forward harmony-gdal-adapter-58b6f98b57-sv5vm -n harmony 5678:5678

5. Submit a harmony client command that will trigger this service.

   • The first time after a restart of the harmony services, you might not have to submit a command because harmony submits a fake request to prime the system and that priming request should be waiting for a debugger to attach.

6. Attach your debugger using a launch.json file like this one

      {
        "name": "Harmony GDAL Adapter Attach",
        "type": "python",
        "request": "attach",
        "connect": {
          "host": "localhost",
          "port": 5678
        },
        "pathMappings": [
          {
          "localRoot": "${workspaceFolder}",
          "remoteRoot": "/home/"
          }
        ]
      }

Contributions:

Pull requests:

Contributions to HGA can be made via submitting a pull request (PR) to this repository. Developers with contributor privileges to the Harmony team within the NASA GitHub organisation should have the ability to create a PR directly within this repository. Other developers will need to follow the fork-and-pull model.

In addition to any DAAC stakeholders, please add members of the EED Data Services team (currently: David Auty, Owen Littlejohns and Matt Savoie) as PR reviewers. One of these developers must approve the PR before it is merged.

NASA GitHub contributers will have the required GitHub permissions to merge their PRs after the peer review is complete. Please consider using the squash-and-merge option when merging a PR to maintain a clean Git history.

The EED Data Services team will merge approved PRs for developers without write access to the HGA repository.

Semantic versioning:

When making changes to the HGA service code, it is important to make updates to version.txt and CHANGE.md. Every time code is merged to the main branch and the merged commits contain changes to version.txt, a Docker image is published to ghcr.io. By only triggering image publication when version.txt is incremented, the existing Docker images for HGA will not be overwritten.

When writing or updating service code, please also update the existing test suite with unit tests to ensure the new functionality performs as expected, and continues to do so following subsequent development.

Deployment of new HGA versions:

After a new Docker image has been published, it will need to be deployed as part of the Harmony Kubernetes cluster. The EED Data Services team will coordinate with Harmony to ensure HGA is updated. If there are specific deployment requirements, such as test data only being available in UAT or time constraints, please communicate these to the Data Services team.

Initial deployment will be to test environments (SIT, then UAT), to allow changes to be tested by all DAACs that are using HGA with their data collections.

It is possible to determine the version of HGA that is deployed to a given Harmony environment via the /versions endpoint, e.g.: https://harmony.earthdata.nasa.gov/versions.