need help
12 TopicsSAP ECC into a Microsoft Fabric Lakehouse using Fabric Data Factory.
We are currently extracting data from SAP ECC into a Microsoft Fabric Lakehouse using Fabric Data Factory. *SAP ECC side:* We are using the SAP .NET Connector (NCo), which communicates with SAP through RFC calls. *Fabric side:* We are using the SAP Table – Application Server connector, which leverages the Open Data Gateway (ODG) to retrieve data from SAP ECC. *Small tables:* Data extraction and loading work successfully with relatively small SAP tables. *Large tables:* When extracting larger tables, the process fails with a "Not enough space" error. *Partitioning / Chunking:* We have tried the partitioning and RFC table options available in the Fabric connector to split large-table extraction into smaller chunks/batches based on indexed fields. *Performance issue:* Although chunking allows us to process the data, the resulting throughput is very low, and extracting large tables takes an excessive amount of time. *Key Challenge* We need to identify a scalable and performant approach for extracting large SAP ECC tables into Microsoft Fabric Lakehouse, while avoiding the "Not enough space" error and significantly improving extraction throughput. In particular, we need to determine whether the bottleneck is related to SAP RFC/NCo processing, Open Data Gateway, Fabric's SAP Table connector, partitioning strategy, batch size, or the underlying data transfer mechanism, and what configuration or architecture would provide better performance for large-volume SAP extractions.Solved66Views0likes5CommentsDAGRun via API
Hi, I am trying to find out how to run Airflow DAG via API call. When I called an API from swagger UI page, it works properly. (using /connection api as example for simplicity) However, if I called the same API from shell script or postman, it will always redirect me to microsoft login page. Is there a way to authenticate using shell script or do I need to change airflow Configuration Overrides to allow basic authentication (if possible)? Is it currently possible to run a DAG from a REST API call using basic authen or is there a way to authenticate using shell script ?10KViews1like3CommentsApache airflow new version and private endpoint support
We have the following questions – if anybody has input it would be much appreciated The current version is 2.10.5 from Feb 10, 2025, current version as of today is 3.2.2 (29 may 2026). Any idea when we may see a more current version and this is something we may change ourself – I cannot see it in for instance settings on the item? According to Fabric roadmap it should be “Q2 2026” (I would like to use our smtp server to send email – this is however connected via. private link service and managed private endpoint so need this). Any idea when Apache airflow will support private networks and expectation of when it is planned for GA?Solved1.9KViews1like3CommentsData Pipeline Parameterization Fails When Using Workspace Variables in REST Activities — Any Workaro
We follow the recommended pattern of using Workspace variables for environment management, but REST activities within Data Pipelines fail to resolve variable references, even though SQL and Notebook activities work fine. The error: Parameter '<workspaceVariableName>' not found. Has anyone successfully used workspace variables inside REST activities? Is this limitation documented somewhere or are we missing a configuration?Solved13KViews2likes3Commentsstuck with MS Learn Tutorial for Apache Airflow Jobs for dbt-fabric Orchestration
Hello, does anyone managed to implement the sample/tutorial Transform data using dbt - Microsoft Fabric | Microsoft Learn ? For us, with a Starter Pool running on Apache Airflow version 2.10.5, only the version of astronomer-cosmos==1.5.1 and dbt-fabric==1.5.0 are valid. ....but when running the DAG 'dags/my_cosmos_dag.py we systematically get a Failed to start error. ThanksSolved7.6KViews1like9CommentsUnauthorized Error in Airflow Task with Microsoft Fabric Plugin (401 Invalid Client)
Hi everyone, I'm encountering an issue while running an Airflow task that interacts with Microsoft Fabric using the apache_airflow_microsoft_fabric_plugin. The task fails with a 401 Unauthorized error, and the logs indicate the following: invalid_client: The request body must contain the following parameter: 'client_assertion' or 'client_secret'. Trace ID: 54844adb-4acd-4720-a99f-8b71ad921100 Correlation ID: fd79265c-dd63-457b-aa31-50ed665fec2d Timestamp: 2024-10-04 12:18:02Z It seems the OAuth2 request is missing either a client_assertion or client_secret. I've verified that my Azure app registration has a client secret configured and that my Airflow connection (fabric-airflow-conn) is set up with the correct credentials. Here's the traceback: [2024-10-04, 14:18:02 CEST] {base.py:73} INFO - Using connection ID 'fabric-airflow-conn' for task execution. [2024-10-04, 14:18:02 CEST] {taskinstance.py:1824} ERROR - Task failed with exception Traceback (most recent call last): File "/home/airflow/.local/lib/python3.8/site-packages/apache_airflow_microsoft_fabric_plugin/hooks/fabric.py", line 132, in _get_token response.raise_for_status() File "/home/airflow/.local/lib/python3.8/site-packages/requests/models.py", line 1021, in raise_for_status raise HTTPError(http_error_msg, response=self) requests.exceptions.HTTPError: 401 Client Error: Unauthorized for url: https://login.microsoftonline.com/117c4e93-3230-40d6-b1cd-57cdee469b2f/oauth2/v2.0/token During handling of the above exception, another exception occurred: Traceback (most recent call last): File "/home/airflow/.local/lib/python3.8/site-packages/apache_airflow_microsoft_fabric_plugin/operators/fabric.py", line 122, in execute self.location = self.hook.run_fabric_item( File "/home/airflow/.local/lib/python3.8/site-packages/apache_airflow_microsoft_fabric_plugin/hooks/fabric.py", line 214, in run_fabric_item headers = self.get_headers() File "/home/airflow/.local/lib/python3.8/site-packages/apache_airflow_microsoft_fabric_plugin/hooks/fabric.py", line 161, in get_headers "Authorization": f"Bearer {self._get_token()}", File "/home/airflow/.local/lib/python3.8/site-packages/apache_airflow_microsoft_fabric_plugin/hooks/fabric.py", line 135, in _get_token raise AirflowException(msg) airflow.exceptions.AirflowException: Response: {"error":"invalid_client","error_description":"AADSTS7000218: The request body must contain the following parameter: 'client_assertion' or 'client_secret'. Trace ID: 54844adb-4acd-4720-a99f-8b71ad921100 Correlation ID: fd79265c-dd63-457b-aa31-50ed665fec2d Timestamp: 2024-10-04 12:18:02Z","error_codes":[7000218],"timestamp":"2024-10-04 12:18:02Z","trace_id":"54844adb-4acd-4720-a99f-8b71ad921100","correlation_id":"fd79265c-dd63-457b-aa31-50ed665fec2d","error_uri":"https://login.microsoftonline.com/error?code=7000218"} Status Code: 401 [2024-10-04, 14:18:02 CEST] {taskinstance.py:1345} INFO - Marking task as FAILED. dag_id=Run_Fabric_Item, task_id=run_fabric_notebook, execution_date=20241004T121800, start_date=20241004T121801, end_date=20241004T121802 [2024-10-04, 14:18:02 CEST] {standard_task_runner.py:104} ERROR - Failed to execute job 104 for task run_fabric_notebook (Response: {"error":"invalid_client","error_description":"AADSTS7000218: The request body must contain the following parameter: 'client_assertion' or 'client_secret'. Trace ID: 54844adb-4acd-4720-a99f-8b71ad921100 Correlation ID: fd79265c-dd63-457b-aa31-50ed665fec2d Timestamp: 2024-10-04 12:18:02Z","error_codes":[7000218],"timestamp":"2024-10-04 12:18:02Z","trace_id":"54844adb-4acd-4720-a99f-8b71ad921100","correlation_id":"fd79265c-dd63-457b-aa31-50ed665fec2d","error_uri":"https://login.microsoftonline.com/error?code=7000218"} Status Code: 401; 50) [2024-10-04, 14:18:02 CEST] {local_task_job_runner.py:225} INFO - Task exited with return code 1 Has anyone else encountered this issue or have insights into how to resolve it? Thanks in advance!Solved7.9KViews0likes3CommentsTransform data using dbt tutorial, not working
After successfully starting the Hello Airflow job example, I am trying a slightly modified version of this tutorial, in a new Airflow job: https://learn.microsoft.com/en-us/fabric/data-factory/apache-airflow-jobs-dbt-fabric The modification is that I am connecting to the default Warehouse with sample data, instead of the mentioned Contoso warehouse. The problem is that the DAG does not run, after trying multiple schedule intervals. Hopefully I have configured the DAG correctly, the only doubt I have is the database field in the profiles YAML file. Should that be the name of the warehouse? EDIT: I just discovered that there is a schedule interval set to daily, which means that the DAG will run at midnight on the day it is started, right? Is this a particulary beginner friendly setting for a tutorial? The tutorial also says to "press play", but this is a scheduled DAG right? EDIT 2: I have set the schedule interval to continuous, and start date two days ago, catchup to False still no DAGs are run. Where can I find the logs? Screenshots:Solved5.4KViews1like5CommentsData workflows - Run a Microsoft Fabric Item Job
Has anyone been able to get a Fabric Data workflow to run a Fabric Item Job? Run a Microsoft Fabric item job in Apache Airflow DAGs. - Microsoft Fabric | Microsoft Learn I have the authentication working, but hitting this error and not sure where to find this module? I cannot find it on PyPI ModuleNotFoundError: No module named 'airflow.providers.microsoft.powerbi'5.9KViews0likes5Comments