Forum Discussion
Creating Fabric Eventstream via API: How to complete the KQL Data Connection configuration?
- 4 months ago
Hi rgsalido ,
as there is currently no official Microsoft documentation or fully supported API approach for configuring the KQL Data connection within Eventstream, we recommend submitting this as feedback or a feature request in the Microsoft Fabric community forum. This will help bring visibility to the gap and allow the product team to consider enhancing API support for this scenario. You can raise your request under the Ideas or Feedback section here: Fabric Ideas - Microsoft Fabric Community
Thanks,
Prashanth
Hi everyone,
Just to close the loop on this thread and share the outcome in case it helps others.
We were able to fully automate the Eventstream → Eventhouse (Direct Ingestion) configuration via the REST API, but there are a few important constraints that are not obvious from the documentation.
In our case, the issue was not that the “Configure KQL Data Connection” step was unavailable via API, but rather that some hidden constraints were not being respected:
- The default stream name must follow the pattern:
{EventstreamName}-stream
This name must be used consistently in the definition (streams and inputNodes).
- The connectionName must be unique per Eventstream when connecting to the same Eventhouse.
- The connectionName must be under 40 characters.
We were generating it using a full GUID, which resulted in values longer than 40 characters. When this limit is exceeded:
- No clear error is returned by the API.
- The dataConnectionId remains all zeros.
- The internal DataConnection is not materialized.
- Ingestion does not start.
After switching to a shorter, unique connectionName (well under 40 characters) and enforcing the correct stream naming pattern, the Eventstream started ingesting data correctly via API, without any manual UI step.
So yes — it is possible to fully automate this scenario, but those constraints must be respected.
Hope this helps someone else facing the same behavior.
Thanks again to the support team for the investigation and guidance.