IDC Health Monitor

Automated health checks for NCI Imaging Data Commons services. Runs daily via GitHub Actions to verify that IDC infrastructure is operational.

Live dashboard: imagingdatacommons.github.io/idc-health-monitor

What's monitored

Service	Test file	Auth required
BigQuery (idc_current.dicom_all)	tests/idc_tests.py	Yes (GCP service account)
IDC API (prod + dev)	tests/idc_tests.py	No
IDC Portal	tests/idc_tests.py	No
DICOMweb public proxy (QIDO-RS, WADO-RS)	tests/test_dicomweb.py	No
DICOMweb Google Healthcare API	tests/test_dicomweb.py	Yes (GCP service account)

Local development

Setup

uv venv
source .venv/bin/activate
uv pip install requests google-cloud-bigquery google-auth

Running tests

Run all tests (requires GCP credentials for BigQuery and Healthcare API tests):

python -m unittest -v tests/idc_tests.py tests/test_dicomweb.py

Run only the DICOMweb proxy tests (no credentials needed):

python -m unittest -v tests.test_dicomweb.TestDICOMwebProxy

GCP authentication for local runs

For tests that require Google Cloud credentials (BigQuery, Healthcare API):

export GOOGLE_APPLICATION_CREDENTIALS=/path/to/service-account.json
export PROJECT_ID=your-gcp-project-id

GitHub Actions

The workflow (.github/workflows/run_tests.yml) runs on:

• Push to main
• Manual dispatch
• Daily schedule (7:30 AM UTC)

Required secrets

Secret	Purpose
BQ_SERVICE_ACCOUNT	GCP service account JSON for BigQuery and Healthcare API access
PROJECT_ID	GCP project ID for BigQuery client

Adding new tests

1. Create a new test file in tests/ following the unittest.TestCase pattern
2. Add docstrings to test methods (displayed on the dashboard)
3. Import and load the new module in scripts/run_tests_json.py
4. Add any new pip dependencies to the workflow's install step

Key constants to maintain

• IDC data version (IDC_VERSION in tests/test_dicomweb.py): Update when IDC releases a new data version. Current: v23.
• Known StudyInstanceUID (KNOWN_STUDY_UID in tests/test_dicomweb.py): Used for targeted DICOMweb queries. Should be a stable, public study unlikely to be removed.

License

MIT