Azure Synapse Analytics known issues
This page lists the known issues in Azure Synapse Analytics, and their resolution date or possible workaround. Before submitting an Azure support request, review this list to see if the issue that you're experiencing is already known and being addressed.
To learn more about Azure Synapse Analytics, see the Azure Synapse Analytics Overview, and What's new in Azure Synapse Analytics?
Active known issues
Azure Synapse Analytics dedicated SQL pool active known issues summary
Customers are unable to monitor their usage of dedicated SQL pool by using metrics
An internal upgrade of our telemetry emission logic, which was meant to enhance the performance and reliability of our telemetry data, caused an unexpected issue that affected some customers' ability to monitor their dedicated SQL pool, tempdb, and Data Warehouse Data IO metrics.
Workaround: Upon identifying the issue, our team took action to identify the root cause and update the configuration in our system. Customers can fix the issue by pausing and resuming their instance, which will restore the normal state of the instance and the telemetry data flow.
Query failure when ingesting a parquet file into a table with AUTO_CREATE_TABLE='ON'
Customers who try to ingest a parquet file into a hash distributed table with AUTO_CREATE_TABLE='ON' can receive the following error:
COPY statement using Parquet and auto create table enabled currently cannot load into hash-distributed tables
Ingestion into an auto-created hash-distributed table using AUTO_CREATE_TABLE is unsupported. Customers that have previously loaded using this unsupported scenario should CTAS their data into a new table and use it in place of the old table.
Queries failing with Data Exfiltration Error
Synapse workspaces created from an existing dedicated SQL Pool report query failure related to Data Exfiltration Protection with generic error message while Data Exfiltration Protection is turned off in Synapse Analytics:
Data exfiltration to '{****}' is blocked. Add destination to allowed list for data exfiltration and try again.
Workaround: If you encountered a similar error, engage Azure Support Team for assistance.
UPDATE STATISTICS failure
Some dedicated SQL Pools can encounter an exception when executing an UPDATE STATISTICS statement. The command results in the message "The provided statistics stream is corrupt" and fails to update your statistics.
When a new constraint is added to a table, a related statistic is created in the distributions. If a clustered index is also created on the table, it must include the same columns (in the same order) as the constraint, otherwise UPDATE STATISTICS commands on those columns might fail.
Workaround: Identify if a constraint and clustered index exist on the table. If so, DROP both the constraint and clustered index. After that, recreate the clustered index and then the constraint ensuring that both include the same columns in the same order. If the table does not have a constraint and clustered index, or if the above step results in the same error, contact the Azure Support Team for assistance.
Tag updates appear to fail
When making a change to the tags of a dedicated SQL pool through Azure portal or other methods, an error message can appear even though the change is made successfully.
Workaround: You can confirm that the change to the tags was successful and ignore/suppress the error message as needed.
Azure Synapse workspace active known issues summary
The following are known issues with the Synapse workspace.
Blob storage linked service with User Assigned Managed Identity (UAMI) is not getting listed
The linked service might not be visible under the Data Hub -> Linked -> Azure Blob Storage after configuring the blob storage linked service to use "User Assigned Managed Identity" authentication in Azure Synapse Analytics.
Workaround: The engineering team is currently aware of this behavior and working on a fix. As an alternative, use "System Assigned Managed Identity" authentication method instead of "User Assigned Managed Identity".
Failed to delete Synapse workspace & Unable to delete virtual network
Deleting a Synapse workspace fails with the error message:
Failed to delete Synapse workspace '[Workspace Name]'. Unable to delete virtual network. The correlationId is ********-****-****-****-************;
Workaround: The problem can be mitigated by retrying the delete operation. The engineering team is aware of this behavior and working on a fix.
REST API PUT operations or ARM/Bicep templates to update network settings fail
When using an ARM template, Bicep template, or direct REST API PUT operation to change the public network access settings and/or firewall rules for a Synapse workspace, the operation can fail.
Workaround: The problem can be mitigated by using a REST API PATCH operation or the Azure portal UI to reverse and retry the desired configuration changes. The engineering team is aware of this behavior and working on a fix.
Known issue incorporating square brackets [] in the value of Tags
In the context of updating tag values within an Azure Synapse workspace, the inclusion of square brackets ([]) will result in an unsuccessful update operation.
Workaround: The current workaround is to abstain from using the square brackets ([]) in Azure Synapse workspace tag values.
No 'GET' API operation dedicated to the "Microsoft.Synapse/workspaces/trustedServiceBypassEnabled" setting
Issue Summary: In Azure Synapse Analytics, there is no dedicated 'GET' API operation for retrieving the state of the "trustedServiceBypassEnabled" setting at the resource scope "Microsoft.Synapse/workspaces/trustedServiceBypassEnabled". While users can set this configuration, they cannot directly retrieve its state via this specific resource scope.
Impact: This limitation impacts Azure Policy definitions, as they cannot enforce a specific state for the "trustedServiceBypassEnabled" setting. Customers are unable to use Azure Policy to deny or manage this configuration.
Workaround: There is no workaround available in Azure Policy to enforce the desired configuration state for this property. However, users can use the 'GET' workspace operation to audit the configuration state for reporting purposes.
This 'GET' workspace operation maps to the 'Microsoft.Synapse/workspaces/trustedServiceBypassEnabled' Azure Policy Alias.
The Azure Policy Alias can be used for managing this property with a Deny Azure Policy Effect if the operation is a PUT request against the Microsoft.Synapse/workspace resource, but it will only function for Audit purposes if the PUT request is being sent directly to the Microsoft.Synapse/workspaces/trustedServiceByPassConfiguration child resource. The parent resource has a property [properties.trustedServiceBypassEnabled] that maps the configuration from the child resource and this is why it can still be audited through the parent resource's Azure Policy Alias.
Since the Microsoft.Synapse/workspaces/trustedServiceByPassConfiguration child resource has no GET operation available, Azure Policy cannot manage these requests, and Azure Policy cannot generate an Azure Policy Alias for it.
Parent Resource: Microsoft.Synapse/workspaces
Child Resource: Microsoft.Synapse/workspaces/trustedServiceByPassConfiguration
The Azure portal makes the PUT request directly to the PUT API for the child resource and therefore the Azure portal, along with any other API requests made outside of the parent Microsoft.Synapse/workspaces APIs, cannot be managed by Azure Policy through a Deny or other actionable Azure Policy Effect.
Azure Synapse Analytics serverless SQL pool active known issues summary
Query failures from serverless SQL pool to Azure Cosmos DB analytical store
Queries from a serverless SQL pool to Azure Cosmos DB analytical store might fail with one of the following error messages:
Resolving CosmosDB path has failed with error 'This request is not authorized to perform this operation'Resolving CosmosDB path has failed with error 'Key not found'
The following conditions must be true to confirm this issue:
- The connection to Azure Cosmos DB analytical store uses a private endpoint.
- Retrying the query succeeds.
Workaround: The engineering team is aware of this behavior and following actions can be taken as quick mitigation:
- Retry the failed query. It will automatically refresh the expired token.
- Disable the private endpoint. Before applying this change, confirm with your security team that it meets your company security policies.
Azure Cosmos DB analytical store view propagates wrong attributes in the column
While using views in Azure Synapse serverless pool over Cosmos DB analytical store, if there is a change on files in the Cosmos DB analytical store, the change does not get propagated correctly to the SELECT statements, the customer is using on the view. As a result, the attributes get incorrectly mapped to a different column in the results.
Workaround: The engineering team is aware of this behavior and following actions can be taken as quick mitigation:
- Recreate the view by renaming the columns.
- Avoid using views if possible.
Alter database-scoped credential fails if credential has been used
Sometimes you might not be able to execute the ALTER DATABASE SCOPED CREDENTIAL query. The root cause of this issue is the credential was cached after its first use making it inaccessible for alteration. The error returned is:
Failed to modify the identity field of the credential '{credential_name}' because the credential is used by an active database file.
Workaround: The engineering team is currently aware of this behavior and is working on a fix. As a workaround you can DROP and CREATE the credentials, which would also mean recreating external tables using the credentials. Alternatively, you can engage Azure Support Team for assistance.
Query failures in serverless SQL pools
Token expiration can lead to errors during their query execution, despite having the necessary permissions for the user over the storage. These error messages can also occur due to common user errors, such as when role-based access control (RBAC) roles are not assigned to the storage account.
Example error messages:
WaitIOCompletion call failed. HRESULT = 0x80070005'. File/External table name: {path}Unable to resolve path '%' Error number 13807, Level 16, State 1, Message "Content of directory on path '%' cannot be listed.Error 16561: External table '<table_name>' is not accessible because content of directory cannot be listed.Error 13822: File {path} cannot be opened because it does not exist or it is used by another process.Error 16536: Cannot bulk load because the file "%ls" could not be opened.
Workaround:
The resolution is different depending on the authentication, Microsoft Entra (formerly Azure Active Directory) or managed service identity (MSI):
For Microsoft Entra token expiration:
For long-running queries, switch to service principal, managed identity, or shared access signature (SAS) instead of using a user identity. For more information, see Control storage account access for serverless SQL pool in Azure Synapse Analytics.
Restart client (SSMS/ADS) to acquire a new token to establish the connection.
For MSI token expiration:
- Deactivate then activate the pool in order to clear the token cache. Engage Azure Support Team for assistance.
Storage access issues due to authorization header being too long
Example error messages in serverless SQL pools:
File {path} cannot be opened because it does not exist or it is used by another process.Content of directory on path {path} cannot be listed.WaitIOCompletion call failed. HRESULT = {code}'. File/External table name: {path}
These generic storage access errors appear when running a query. The issue might occur for a user in one workspace but would work properly in other workspaces. This behavior is expected due to token size.
Check the Microsoft Entra token length by running the following command in PowerShell. If the token length is close to 11000 or longer, see Mitigation section.
(Get-AzAccessToken -ResourceUrl https://database.chinacloudapi.cn).Token.Length
Workaround:
Suggested workarounds are:
- Switch to Managed Identity storage authorization as described in the storage access control.
- Decrease number of security groups (having 90 or fewer security groups results with a token that is of compatible length).
- Increase number of security groups over 200 (as that changes how token is constructed, it will contain an MS Graph API URI instead of a full list of groups). It could be achieved by adding dummy/artificial groups by following managed groups, after you would need to add users to newly created groups.
Querying a view shows unexpected results
When you query the view for which the underlying schema has changed after the view was created, you may encounter unexpected results. This means that the view references columns or objects that were modified or no longer exist. To overcome this you need to manually adjust the view definition to align with the underlying schema changes.
Workaround: Manually adjust the view definition.
Azure Synapse Analytics Apache Spark pool active known issues summary
The following are known issues with the Synapse Spark.
Query failure with a LIKE clause using Synapse Dedicated SQL Pool Connector in Spark 3.4 runtime
The open source Apache Spark 3.4 has introduced an issue, which escapes special characters, but Synapse SQL does not support the escape keyword. When customers use the Azure Synapse Dedicated SQL Pool Connector for Apache Spark, it can generate an invalid SQL query for Synapse SQL and the Synapse Spark notebook or batch job would throw an error similar to:
com.microsoft.spark.sqlanalytics.SQLAnalyticsConnectorException: com.microsoft.sqlserver.jdbc.SQLServerException: Parse error at line: 1, column: XXX: Incorrect syntax near ''%test%''
Workaround: The engineering team is currently aware of this behavior and working on a fix. If you encountered a similar error, please engage Microsoft Support Team for assistance and to provide a temporary workaround.
Recently closed known issues
| Synapse Component | Issue | Status | Date Resolved |
|---|---|---|---|
| Azure Synapse serverless SQL pool | Queries using Microsoft Entra authentication fails after 1 hour | Resolved | August 2023 |
| Azure Synapse serverless SQL pool | Query failures while reading Cosmos DB data using OPENROWSET | Resolved | March 2023 |
| Azure Synapse Apache Spark pool | Failed to write to SQL Dedicated Pool from Synapse Spark using Azure Synapse dedicated SQL pool Connector for Apache Spark when using notebooks in pipelines | Resolved | June 2023 |
| Azure Synapse Apache Spark pool | Certain spark job or task fails too early with Error Code 503 due to storage account throttling | Resolved | November 2023 |
Azure Synapse Analytics serverless SQL pool recently closed known issues summary
Queries using Microsoft Entra authentication fails after 1 hour
SQL connections using Microsoft Entra authentication that remain active for more than 1 hour starts to fail. This includes querying storage using Microsoft Entra pass-through authentication and statements that interact with Microsoft Entra ID, like CREATE EXTERNAL PROVIDER. This affects every tool that keeps connections active, like query editor in SSMS and ADS. Tools that open new connection to execute queries aren't affected, like Synapse Studio.
Status: Resolved
Query failures while reading Azure Cosmos DB data using OPENROWSET
Queries from serverless SQL pool to Cosmos DB Analytical Store using OPENROWSET fails with the following error message:
Resolving CosmosDB path has failed with error 'bad allocation'.
Status: Resolved
Azure Synapse Analytics Apache Spark pool recently closed known issues summary
Failed to write to SQL Dedicated Pool from Synapse Spark using Azure Synapse dedicated SQL pool connector for Apache Spark when using notebooks in pipelines
While using Azure Synapse dedicated SQL pool Connector for Apache Spark to write Azure Synapse Dedicated pool using Notebooks in pipelines, we would see an error message:
com.microsoft.spark.sqlanalytics.SQLAnalyticsConnectorException: COPY statement input file schema discovery failed: Cannot bulk load. The file does not exist or you don't have file access rights.
Status: Resolved
Certain spark job or task fails too early with Error Code 503 due to storage account throttling
Between October 3, 2023 and November 16, 2023, few Azure Synapse Analytics Apache Spark pools could have experienced spark job/task failures due to storage API limit threshold being exceeded.
Status: Resolved