Forum Discussion
Does MashupDefaultCommandTimeout affect Snowflake large-result failures?
Hi everyone,
We are troubleshooting an intermittent Power BI Service refresh issue with Snowflake through an On-premises Data Gateway.
Our architecture is:
Power BI Service → On-premises Data Gateway → Corporate Network/Proxy/Firewall → Snowflake
The issue occurs mainly with large datasets. The Snowflake query completes successfully, but the refresh sometimes fails while retrieving the result data.
The errors include:
An existing connection was forcibly closed by the remote host wsarecv: An existing connection was forcibly closed by the remote host CURLcode=35 SSL connect error Arrow chunk download failed – max retries done
The same report can fail on one refresh and succeed on retry.
We also tested multiple reports simultaneously. In one test, 11 reports were refreshed at the same time: 8 succeeded and 3 failed. The 3 failed reports succeeded after 1–2 retries.
Snowflake Support confirmed that the query itself is completing and that the failure appears to occur while retrieving the large result set from their STAGE/S3 endpoints.
We have also tested Snowflake Connector Implementation 1.0 instead of 2.0, but the same intermittent failures occurred.
My question
Got a suggestion to add the following setting to:
Microsoft.PowerBI.DataMovement.Pipeline.GatewayCore.dll.config
<setting name="MashupDefaultCommandTimeout" serializeAs="String"> <value>1800000</value> </setting>
This sets the timeout to 30 minutes, followed by a Gateway service restart.
Would MashupDefaultCommandTimeout actually affect this type of failure, where the Snowflake query completes successfully but the connection is reset while retrieving/downloading a large result set?
Or does this timeout only apply to the Mashup/command execution phase, meaning it would not help with the subsequent Snowflake/S3 result download?
If it does not apply to the download phase, what Gateway-side setting or diagnostic would you recommend checking for this type of intermittent wsarecv / CURLcode=35 failure?
We are also checking with our Network Team for:
- TLS/SSL inspection
- Proxy/firewall session or idle timeouts
- Concurrent connection limits
- NAT/PAT exhaustion
- Bandwidth throttling
Any guidance from Power BI Gateway/experts on whether the 30-minute Mashup timeout is relevant to this specific scenario would be greatly appreciated.
Thanks!
1 Reply
- v-achippaCommunity Support
Hi manoj_0911,
Thank you for reaching out to Microsoft Fabric Community.
The MashupDefaultCommandTimeout setting will not resolve this issue.
The Snowflake query completes successfully, but the connection is interrupted while the gateway downloads the large result set from snowflake’s stage or storage endpoints.
Please work with your network team to check for TLS/SSL inspection or proxy/firewall session limits affecting the snowflake/S3 connection and allow uninterrupted HTTPS access from the gateway server to all snowflake and stage or storage endpoints.
Thanks and regards,
Anjan Kumar Chippa