Bulk-import Jpa-Server-Starter ID constraint error

[Loiotile]

Most of the times we try to run a bulk-import on our unsorted data (resources kept inside .ndjson files, with multiple references to eachother ids that need to be maintained), we encounter the following error: "ERROR c.u.f.c.BulkImportCommand HTTP 500 : Job is in FAILED state with 22 error count. Last error: Too many errors: 4. Last error msg was HAPI-0550: HAPI-0825: The operation has failed with a client-assigned ID constraint failure. This typically means that multiple client threads are trying to create a new resource with the same client-assigned ID at the same time, and this thread was chosen to be rejected. It can also happen when a request disables the Upsert Existence Check". This error doesn't occur for smaller groups of data; also, in rare occasions the program run with same settings and data causing this error can end up successful.
In our application.yaml have "autocreate_placeholder_reference_targets: true", in order to create placeholders for resources referenced before the resources themselves are actually populated, and we assumed that these placeholders could have been the reason behind the error.

• HAPI FHIR Version 5
• OS: Windows
• Browser: Microsoft Edge

Additional context:
Use of hapi-fhir-CLI with localhost and PostgreSQL

[jkiddo]

I have the same observation when trying to do even small synthea-based imports running on H2.