README.md

FHIR to Server

Reads FHIR Bundles from a Kafka topic and sends them to a FHIR server

Works great with the MIRACUM Stream Processors chart!

Container

Published as ghcr.io/miracum/kafka-fhir-to-server.

Configuration

Environment variable	Description	Default
BOOTSTRAP_SERVERS	List of Kafka Bootstrap servers	""
TOPIC	Kafka topic(s) to read FHIR resources from	fhir-msg
FHIR_FILTER_EXPRESSION	FHIR Path expression to filter resources by. Must return a boolean result.	""
FHIR_OVERRIDE_BUNDLE_TYPE_WITH	A FHIR Bundle type to override the type of any received bundle with.	""
FHIR_AUTH_BASIC_ENABLED	Enable HTTP Basic Auth when interacting with the FHIR server.	false
FHIR_AUTH_BASIC_USERNAME	HTTP Basic Auth username for the FHIR server.	""
FHIR_AUTH_BASIC_PASSWORD	HTTP Basic Auth password for the FHIR server.	""
FHIR_HTTP_TIMEOUT_SECONDS	HTTP client timeout in seconds when interacting with the FHIR server	60
FHIR_MERGE_BATCHES_INTO_SINGLE_BUNDLE_ENABLED	Enable merging bundles read as a batch from the input topic into a single topic composed of all individual resources.	false
FHIR_MERGE_BATCHES_INTO_SINGLE_BUNDLE_ENTRY_UNIQUENESS_FHIRPATH_EXPRESSION	A FHIRPath expression evaluated against each bundle.entry. The resulting string represents the resource identity. If multiple entries have the same identity, only the one from the most recently received message is used.	"resource.id.toString()"
FHIR_MERGE_BATCHES_INTO_SINGLE_BUNDLE_BUNDLE_MAX_SIZE	If set, the bundles to be sent to the server are first partitioned into several bundles containing at most this settings resources. Useful if the potential total size of a merged bundle may exceed the limit supported by the server.	null

See application.yml for more options.

Sending resources to S3-compatible object storage (Experimental)

Environment variable	Description	Default
S3_ENABLED	Set to true to persist resources as ndjson in object storage instead of sending them to a FHIR server.	false
S3_ENDPOINT_URL	Object storage endpoint url	""
S3_ACCESS_KEY	The access key. Can also be left empty to use the AWS_ACCESS_KEY_ID environment variable instead.	""
S3_SECRET_KEY	The secret key. Can also be left empty to use the AWS_SECRET_ACCESS_KEY environment variable instead.	""
S3_BUCKET_NAME	The name of the bucket to store the resources. The actual resources are grouped by their type and stored using the current epoch timestamp, e.g. <S3_BUCKET_NAME>/Patient/bundle-123456.ndjson	"fhir"
S3_OBJECT_NAME_PREFIX	An optional prefix to prepend to the object name: <S3_BUCKET_NAME><S3_OBJECT_NAME_PREFIX>Patient/bundle-123456.ndjson	""

Observability

The application publishes useful Prometheus metrics at /actuator/prometheus.

Development

Setup Kafka

docker compose up

Run the app

./gradlew :bootRun

Kubernetes

Create a local KinD cluster

kind create cluster

Prerequisite: Install Strimzi & Kafka

helm repo add strimzi https://strimzi.io/charts/
helm install strimzi-kafka-operator strimzi/strimzi-kafka-operator
kubectl apply -f k8s/kafka-cluster.yaml

Install a FHIR server and launch the fhir-to-server deployment

skaffold dev