Forum Discussion
Snowflake Mirroring – Error 100089 / AuthenticationError during replication
Hi everyone,
I'm experiencing an issue with Microsoft Fabric Mirroring for Snowflake.
On the Snowflake side, the COPY INTO command issued by Fabric Mirroring fails with the following error:
Error 100089: Failed to access remote file: access denied
On the Fabric side, the mirrored database shows AuthenticationError and the status "Running with warnings."
Interestingly, some large tables (around 700 million rows) successfully completed the initial replication, while other tables fail with the error above.
In addition, currently all tables show the following warning:
Could not obtain a connection from the pool within a given timeout
ErrorCode: AuthenticationError
SequenceNumber: 154
Since some very large tables were replicated successfully, I'm not sure whether this is actually related to Snowflake permissions/authentication, or whether it could be related to the connection pool, temporary staging/remote file access, or Fabric Mirroring itself.
Has anyone encountered a similar issue with Snowflake Mirroring?
In particular, I would appreciate any guidance on:
- What could cause Failed to access remote file: access denied for the COPY INTO command generated by Fabric Mirroring?
- Why would only some tables fail while others successfully complete the initial replication?
- Could the Could not obtain a connection from the pool within a given timeout warning be related to the Snowflake COPY INTO access-denied error?
- Are there any logs or additional diagnostics in Fabric that can help identify the root cause?
Any suggestions would be greatly appreciated. Thanks!
Hi erina_1209,
Thanks for reaching out to the Microsoft Fabric Community forum.
Based on Microsoft’s Snowflake Mirroring documentation, a few areas may be worth checking:
• PREVENT_UNLOAD_TO_INLINE_URL : Fabric normally stages Snowflake data through an inline URL. If this Snowflake account-level parameter is set to true, Mirroring falls back to direct reads for public-endpoint connections. Microsoft notes that this can result in slower replication and an increased risk of connection timeouts, particularly for large datasets.
• Snowflake permissions : Verify that the Mirroring user has the required CREATE STREAM, SELECT table, SHOW tables, and DESCRIBE tables permissions.
• Table types : Compare the failing tables with those that replicated successfully. Managed and Iceberg tables are supported, while external, transient, temporary, and dynamic tables aren't supported.
• Monitoring/logs : Fabric’s Mirroring monitoring provides database/table-level status, rows replicated, and last completed time. Workspace monitoring provides more detailed execution information, including failures, processed rows/bytes, latency, and error messages through MirroredDatabaseTableExecution.
The connection-pool timeout is worth investigating alongside the connectivity and staging configuration, particularly if PREVENT_UNLOAD_TO_INLINE_URL is enabled.
I would therefore check the above configuration and the detailed Mirroring logs to narrow down whether the issue is related to connectivity, permissions, or a particular table.
For more details, please refer to the below official Microsoft documentation:
Microsoft Fabric Mirrored Databases From Snowflake - Microsoft Fabric | Microsoft Learn
Limitations for Microsoft Fabric Mirrored Databases From Snowflake - Microsoft Fabric | Microsoft Learn
Monitor Mirrored Database Replication - Microsoft Fabric | Microsoft Learn
Mirrored database operation logs - Microsoft Fabric | Microsoft Learn
I hope this helps. Please feel free to reach out if you have any further questions.
Thank you.
3 Replies
v-abhinavmuCommunity Support
Hi erina_1209,
Thanks for reaching out to the Microsoft Fabric Community forum.
Based on Microsoft’s Snowflake Mirroring documentation, a few areas may be worth checking:
• PREVENT_UNLOAD_TO_INLINE_URL : Fabric normally stages Snowflake data through an inline URL. If this Snowflake account-level parameter is set to true, Mirroring falls back to direct reads for public-endpoint connections. Microsoft notes that this can result in slower replication and an increased risk of connection timeouts, particularly for large datasets.
• Snowflake permissions : Verify that the Mirroring user has the required CREATE STREAM, SELECT table, SHOW tables, and DESCRIBE tables permissions.
• Table types : Compare the failing tables with those that replicated successfully. Managed and Iceberg tables are supported, while external, transient, temporary, and dynamic tables aren't supported.
• Monitoring/logs : Fabric’s Mirroring monitoring provides database/table-level status, rows replicated, and last completed time. Workspace monitoring provides more detailed execution information, including failures, processed rows/bytes, latency, and error messages through MirroredDatabaseTableExecution.
The connection-pool timeout is worth investigating alongside the connectivity and staging configuration, particularly if PREVENT_UNLOAD_TO_INLINE_URL is enabled.
I would therefore check the above configuration and the detailed Mirroring logs to narrow down whether the issue is related to connectivity, permissions, or a particular table.
For more details, please refer to the below official Microsoft documentation:
Microsoft Fabric Mirrored Databases From Snowflake - Microsoft Fabric | Microsoft Learn
Limitations for Microsoft Fabric Mirrored Databases From Snowflake - Microsoft Fabric | Microsoft Learn
Monitor Mirrored Database Replication - Microsoft Fabric | Microsoft Learn
Mirrored database operation logs - Microsoft Fabric | Microsoft Learn
I hope this helps. Please feel free to reach out if you have any further questions.
Thank you.- v-abhinavmu
Hi erina_1209,
I wanted to check if you had the opportunity to review the information provided. Please feel free to contact us if you have any further questions.
Thank you. - v-abhinavmu
Hi erina_1209,
May I check if this issue has been resolved? If not, Please feel free to contact us if you have any further questions.
Thank you