Forum Discussion
MetadataEvaluationGatewayTimeout
Hi everyone,
I’d like to share some updates regarding my case.
After analyzing the gateway logs, I identified high CPU and RAM usage on one of the gateways in the cluster. This happens because the cluster is shared with other departments, leading to a high volume of simultaneous requests. Based on the documentation, we might need to add another gateway to the cluster.
Also, support assisted me with a Dataflow Gen2 issue. It is necessary to deploy Dataflow Gen2 using CI/CD across all workspaces or specific folders. This ensures updates are correctly applied. Microsoft is currently working on this.
As a result, my Dataflows Gen2 are now running better.
Thanks.
Hi elissandrorosa,
MetadataEvaluationGatewayTimeout is raised while Power Query Online is asking the gateway to evaluate metadata (schema, navigation, and validation) for your query. With SAP BW, those metadata calls can be slower or bursty (cube navigation, hierarchies, key figure resolution), so the gateway session hits a timeout before the actual data read even starts. That’s why Gen1 might “work” while Gen2 intermittently fails-Gen2 runs in a different service path and tends to surface the metadata phase more explicitly.
Some troubleshooting steps
- Update everything: Upgrade the on-premises data gateway to the latest version and make sure the SAP NetWeaver RFC libraries (64-bit) are installed on the gateway host and match your SAP patch level.
- Reduce metadata round-trips in your M: navigate directly to the InfoProvider/cube and explicitly set the table type and column types (avoid “browsing” steps that expand many folders/levels). Hard-typing columns often prevents extra schema probes.
- Tune the gateway (admin): increase the Mashup container working set on the gateway and avoid memory pressure.
Gateway performance and tuning - Network path: ensure the gateway host has a low-latency, stable route to SAP BW (no flaky VPN, SSL inspection, or idle TCP timeouts on load balancers).
- Retry in Gen2 with a simpler test query that returns a tiny slice (restrict to one InfoProvider, one period) to confirm the path is stable, then add transforms incrementally.
If it still flaps, grab the Activity ID / Correlation ID and open a Fabric support ticket-Microsoft can see the server-side timeout boundary for that request.
Open a Fabric support ticket
If you found this helpful, consider giving some Kudos. If I answered your question or solved your problem, mark this post as the solution.
Thanks tayloramy
I'll evaluate this point to minimize the problem. Thank you again for your information.
Hi elissandrorosa ,
Thanks for the update, Glad to know you’ll be checking those points. Please let us know how it goes after you make the changes.
If the issue still happens, try to open a Fabric support ticket so the team can look into the server-side details.
Best Regards,
Community Support Team.Hi elissandrorosa ,
We are following up once again regarding your query. Could you please confirm if the issue has been resolved through the support ticket with Microsoft?
If the issue has been resolved, we kindly request you to share the resolution or key insights here to help others in the community.
Thank you.
Hello!
We'll be migrating capacity soon, and our IT team will then provide support through a Microsoft ticket.
Thanks.
