Power BI Gateway - The integration runtime is busy now

[JVDA98]

I am getting since last week the following error for an on-premise data gateway when I try to preview data in data pipelines.

The scheduled pipelines don't fail hard, but they time-out after x hours.

"The integration runtime is busy now. Please retry the operation later."

The error is a bit weird as there is no integration runtime on the server or used within Fabric.

A restart of the gateway resolves it for couple of days, but then it comes back...

Gateway has the latest version  installed.

Kind regards,
Jarne

[JVDA98 · Accepted solution]

So after

- updating the Power BI Gateway to the latest version

- Restarting the Gateway Service

- Restarting the Gateway Update Service

- Restarting the Server 

The issue is gone (for now).

Thanks for the responses.

[lbendlin]

If you have a Pro license you can open a Pro ticket at https://admin.powerplatform.microsoft.com/newsupportticket/powerbi
Otherwise you can raise an issue at https://community.fabric.microsoft.com/t5/Issues/idb-p/Issues .

[v-dineshya]

Hi @JVDA98 ,

As mentioned by @lbendlin , Have you raised a support ticket regarding your issue?  If you have already done so, we kindly request you to share your feedback regarding the issue raised.In case a solution has been provided, we would be grateful if you could share it with the community. This will assist others facing similar challenges and benefit the wider community. If you need any more assistance, please feel free to connect with the Microsoft Fabric community.

Regards,

Dinesh

[v-dineshya]

Hi @JVDA98 ,

Thank you for reaching out to the Microsoft Community Forum.

Please try below things to fix the issue.

1. Restart the gateway service and the machine.

2. Update to the latest gateway build (December 2025 is 3000.298).

3. Reduce concurrent pipeline activity while testing. IR errors spike when many Lookups/copy activities test/browse connectors at once. Try a single-activity pipeline or stagger schedules to see if the error disappears.

4. On heavy preview/browse or large copy, the node can get CPU or memory starved, leading to timeouts/busy. adding CPU/RAM or moving to another node with more resources. Aim for dedicated VM, not a shared app host.

5. Multiple nodes provide HA and spread load; swapping the connector’s node has resolved this exact symptom for others. If you already have a cluster, temporarily disable a “sick” node and bind the connector to a healthier one.

6. Missing/undersized JRE on OPDG can cause JNI failures and “busy” behavior especially during schema inference or file parsing. Ensure _JAVA_OPTIONS sets reasonable heap for large files and reboot the node.

7. Look for AsyncOperationStreamingCancelledException or gateway pipeline cancellations, those point to downstream slowness or resource limits. Gather logs from the gateway UI or a diagnostics script to see trends.

8. Check Fabric capacity & region service health.

9. Try moving the workload to off-peak windows or split activities.
Large previews/browse of file shares or big lookups can wedge IR threads. Filter columns, avoid browsing huge folders via the UI, and prefer parameterized paths in Copy activities.

Please refer below links.

Troubleshoot on-premises data gateways - Power BI | Microsoft Learn

On-premises data gateway December 2025 release | Microsoft Fabric Blog | Microsoft Fabric

Solved: The Data Factory runtime is busy now. Please retry... - Microsoft Fabric Community

Troubleshoot connectors - Microsoft Fabric | Microsoft Learn

Solved: On-Prem data gateway Timeout / Cancel errors on mo... - Microsoft Fabric Community

Solved: Re: Error Code: 9518 - Microsoft Fabric Community

I hope this information helps. Please do let us know if you have any further queries.

Regards,

Dinesh

[JVDA98]

Additional information:

Gateway is the latest version

Multiple restarts already regarding the gateway & server

Gateway is only used for one pipeline, the one that is failing

On quite moments, it gives the same issue

No other Integration Runtimes / ... is preset on the server.

Fabric capacity is only at 50%

No complex calculations is done in the pipelines, also not "big data", just pure import of the raw tables in the source.