Microsoft has formally confirmed a critical flaw in recent Windows updates that causes severe database performance degradation. Enterprise environments running SQL Server may experience escalating resource exhaustion due to persistent, unreleased database connections.
The issue stems from cumulative updates released on October 13 and subsequent cumulative packages. When database servers run queries against external data sources, active sessions fail to terminate, leaving idle connections open indefinitely.
windows update sql server connection leak bug
Microsoft confirmed that recent cumulative updates disrupt connection management in SQL Server environments when executing queries through linked servers. The defect causes connections established with external database engines to remain active long after queries finish execution, leading to server resource exhaustion over time.
Microsoft Investigates SQL Server Performance Degradation
According to Microsoft's health dashboard advisory, the underlying defect disrupts the typical session lifecycle within the database engine. When SQL Server executes a query against a remote host, it opens a network path, processes the required data, and is expected to tear down the connection. Under the affected updates, the closing step fails silently.
As applications continue issuing requests throughout production cycles, ghost connections accumulate on host systems. Over extended operational windows, this accumulation consumes vital system handles, memory overhead, and network ports. Administrators monitoring affected hosts report unexpected performance drops, increased query latency, and eventual connection timeouts across dependent enterprise applications.
The issue comes alongside other recent system management challenges, such as when Microsoft resolved Windows 11 installation failures in cumulative update releases, illustrating the complex interactions between operating system patches and core platform components.
Linked Server Queries and Connection Accumulation Details
Technical details published by Microsoft indicate that the flaw specifically impacts SQL Server linked server configurations utilizing Object Linking and Embedding Database (OLE DB) providers. Linked servers allow SQL Server to execute distributed queries against heterogeneous data sources outside the local instance.
Under normal conditions, OLE DB connection pooling handles session allocation and teardown dynamically. However, changes introduced in the October updates prevent the provider interface from releasing target sockets. Each time a SQL Agent job or user session queries an external system, an additional idle session is pinned to the target database. Microsoft noted that connections to non-Microsoft platforms, including Oracle and PostgreSQL databases, are heavily affected by this behavior.
In enterprise environments with automated background jobs running every few minutes, hosts can accumulate thousands of orphan connections within a single operational shift, quickly pushing external database engines to their maximum connection caps.
Impact on External Databases and Affected Windows Releases
The scope of affected operating systems is extensive, spanning virtually all actively serviced client and server editions. On the client side, affected platforms include Windows 10 Enterprise LTSC releases alongside Windows 11 versions prior to 26H2. On the server side, the vulnerability affects Windows Server 2012, 2012 R2, 2016, 2019, 2022, and Windows Server 2025.
Systems running standalone SQL Server instances without linked server definitions or those querying only local databases are unaffected by this specific defect. However, modern hybrid corporate architectures frequently rely on linked server endpoints to bridge reporting structures between SQL Server and external database infrastructure.
The connection leaks add to a growing list of administrative headaches during recent patching cycles. For instance, sysadmins recently had to navigate stricter execution rules when Microsoft tightened PowerShell script signing rules, requiring additional maintenance planning across corporate IT environments.
Temporary Workarounds and Official Fix Timeline
Microsoft has announced plans to issue an out-of-band (OOB) emergency update within days to address the root cause across all impacted Windows releases. In the interim, official guidance offers limited temporary workarounds, none of which represent a permanent resolution.
Database administrators can manually terminate lingering idle connections on target database platforms or restart the main SQL Server service to release socket handles. However, Microsoft explicitly warned that killing active processes or cycling database services carries operational risk. Forcefully dropping connections can abort active transactions, cause data inconsistency, or disrupt connected user sessions. Furthermore, restarting services offers only temporary relief, as connection accumulation resumes immediately as new queries execute.
For mission-critical production environments where service restarts are unacceptable, Microsoft advises IT managers to contact Microsoft Support for Business to receive an interim enterprise mitigation package. Systems administrators are advised to monitor open session metrics on database gateways until the permanent out-of-band patch is deployed.
