Configuration

Bunny can be configured using environment variables. If you’re running Bunny as part of a composed stack of containers, these are set in the compose.yml.

services:
  bunny:
    image: ghcr.io/health-informatics-uon/hutch/bunny:edge
    environment:
      - DATASOURCE_DB_USERNAME=postgres
      - DATASOURCE_DB_PASSWORD=postgres
      - DATASOURCE_DB_DATABASE=postgres
      - DATASOURCE_DB_DRIVERNAME=postgresql
      - DATASOURCE_DB_SCHEMA=public
      - DATASOURCE_DB_PORT=5432
      - DATASOURCE_DB_HOST=db
      - TASK_API_BASE_URL=http://relay:8080/link_connector_api
      - TASK_API_USERNAME=username
      - TASK_API_PASSWORD=password
      - TASK_API_TYPE=a
      - COLLECTION_ID=collection_id
      - LOW_NUMBER_SUPPRESSION_THRESHOLD=10
      - ROUNDING_TARGET=10

Task API

To configure Bunny to communicate with a Task API, such as Relay or RQuest, the TASK_API variables must be set.

Host examples:

• If Bunny and Relay are running in the same Docker Compose stack, the host should be the Relay service name in the stack, e.g. http://relay
• If you have Relay running locally, and Bunny running locally, the host should be http://localhost
• If you have any Task API running elsewhere, the host should be e.g. https://my-task-api.com with the correct hostname
• The Base URL should include /link_connector_api at the end of the path
• Requests from Bunny are authenticated by the TASK_API_USERNAME and TASK_API_PASSWORD, please put your Task API credentials here

Task Type

If you would like Bunny to connect directly with RQUEST, you need to add TASK_API_TYPE.

This can only be either a (Availability), b (Distribution and PHEWAS) or c (AnalyticsGwas, AnalyticsGwasQuantitiveTrait, AnalyticsBurdenTest).

Further guidance is available in the Connect to RQuest How To Guide.

Database

The database connection is configured by the variables prefixed with DATASOURCE_

Managed Identity

When DATASOURCE_USE_AZURE_MANAGED_IDENTITY is enabled, the application will connect to Azure SQL Server using Azure Managed Identity instead of traditional credentials. This allows the service to authenticate securely without requiring a stored username or password. If you are using a user-assigned identity, you can provide its client ID through DATASOURCE_AZURE_MANAGED_IDENTITY_CLIENT_ID.

DuckDB

Trino

Bunny can connect to databases Trino, though this functionality is untested.

Disclosure Control

• Low number suppression defaults to ON with a value of 10 such that any result with a count below 10 will return 0.
• Rounding defaults to ON with a value of 10 such that results are returned to the nearest 10.

Read more about disclosure controls.

Polling

Bunny continually polls upstream to a Task API for new tasks, these settings will configure that functionality.

Distribution Tasks Cache

Bunny can cache the results of a distribution task when it starts, and respond to those queries with the cached result. This is useful for large OMOP databases, where distribution tasks can take longer.

These settings can configure the specifics of this functionality.

Logging

OpenTelemetry

Bunny can use OpenTelemetry to export traces and metrics to your chosen collector. You will need to either have your own observability backend setup (for traces, metrics, and logs) or modify/use our minimal implementation (Loki, Tempo, Prometheus, Grafana) specified in the observability.compose.yml.