Initiate and monitor a FHIR Bulk Data $import operation to asynchronously ingest NDJSON resources

Verified steps

1. Confirm the FHIR server supports bulk import by checking the CapabilityStatement or server documentation; $import is not yet part of the ratified Bulk Data IG (as of the v3.0.0 ballot) but is defined in server-specific or draft proposals — confirm the operation definition URL the server expects
2. Construct the $import request body as a JSON object listing the input NDJSON files: include 'resourceType' to import (e.g., 'Patient', 'Observation') and the storage URL (e.g., an S3 or Azure Blob presigned URL) where the NDJSON files are accessible
3. POST the $import request to [base]/$import (system-level) or a server-defined endpoint with the header 'Prefer: respond-async'; the server returns HTTP 202 Accepted with a Content-Location header pointing to a status polling URL
4. Poll the Content-Location URL with GET requests; the response includes X-Progress (percentage or status text) until the operation completes; on completion the server returns HTTP 200 with a manifest listing imported resource counts and any error files
5. Download and inspect the error NDJSON file (if provided in the manifest) to identify individual resources that failed to import due to validation errors, duplicate conflicts, or referential integrity issues
6. After successful import, verify a sample of imported resources are retrievable via standard FHIR read or search operations to confirm the data was correctly stored