transient-connection-errors.md

author	dlevy-msft-sql
ms.author	dlevy
ms.service	sql
ms.topic	include
ms.date	06/22/2026
ai-usage	ai-assisted

The following errors are transient when they occur during connection establishment or while sending a request to the server. Retry on a short, bounded backoff. Errors that persist past a few retries usually indicate a configuration problem (wrong server, missing permissions, exhausted quota) that retry won't fix.

Error	Message	Troubleshooting
64	A connection was successfully established with the server, but then an error occurred during the login process. (provider: TCP Provider, error: 0 - The specified network name is no longer available.)	The TCP connection drops mid-handshake. Not a credential failure. If it persists, check for client-side network instability or an intermediate device that drops half-established connections.
233	The client was unable to establish a connection because of an error during connection initialization process before login.	Pre-login transport or TLS failure. The server commonly returns it when it can't accept the connection (resource exhaustion, max-connections reached, or an unsupported client). Not a credential failure. Verify server health, then check the client login timeout, TLS settings, and client/server TLS version compatibility.
4060	Cannot open database "%.*ls" requested by the login. The login failed.	The login authenticates but can't open the requested database. Transient causes include the database being in transition (failover, restore, scaling) or auto-paused. Persistent causes (database doesn't exist, login lacks access) won't be fixed by retry; check the database name, login mapping, and database state.
4221	Login to read-secondary failed due to long wait on 'HADR_DATABASE_WAIT_FOR_TRANSITION_TO_VERSIONING'.	The replica isn't available for login because row versions are missing for transactions that were in flight when the replica was recycled. Roll back or commit the active transactions on the primary to resolve the issue. Mitigate by avoiding long write transactions on the primary.
10053	A transport-level error has occurred when sending the request to the server. (provider: TCP Provider, error: 0 - An established connection was aborted by the software in your host machine.)	The local side aborts the connection. Check client-side network health and any local firewall or VPN client.
10054	A transport-level error has occurred when sending the request to the server. (provider: TCP Provider, error: 0 - An existing connection was forcibly closed by the remote host.)	The remote side sends a TCP reset. Common causes: the peer process crashed, a firewall injected a reset, or the Azure SQL gateway closed an idle connection. For idle-reset patterns, enable TCP keepalive on the client or shorten the connection-pool idle timeout.
10928	Resource ID: %d. The %s limit for the database is %d and has been reached. See 'http://go.microsoft.com/fwlink/?LinkId=267637' for assistance.	The database exceeds an Azure SQL resource governance limit. Resource ID 1 indicates the worker limit; Resource ID 2 indicates the session limit. Identify the limit type from the message, then reduce concurrency, scale up the database, or shorten long-running operations holding the resource.
10929	Resource ID: %d. The %s minimum guarantee is %d, maximum limit is %d, and the current usage for the database is %d. However, the server is currently too busy to support requests greater than %d for this database.	The database is over its minimum guarantee and the underlying server is throttling. Retry typically succeeds when neighbor load drops. Sustained occurrences indicate you need a higher service tier or a less noisy environment.
40020, 40143, 40166, 40540	Reported in the Error code %d slot of error 40197 during failover.	Sub-codes embedded in a 40197 failover message that some paths surface as the top-level error number. Treat them the same as 40197.
40197	The service has encountered an error processing your request. Please try again. Error code %d.	A software upgrade, hardware failure, or other failover event in Azure SQL. Reconnecting routes you to a healthy replica. The embedded error code identifies the failover type. If the error persists, capture the session tracing ID and contact support.
40501	The service is currently busy. Retry the request after 10 seconds. Incident ID: %ls. Code: %d.	Azure SQL engine throttling. The recommended floor is a 10-second backoff. Sustained throttling indicates the workload has exceeded the database's resource allocation; scale up the service tier or reduce concurrency.
40613	Database '%.*ls' on server '%.*ls' is not currently available. Please retry the connection later. If the problem persists, contact customer support, and provide them with the session tracing ID of '%.*ls'.	The database is unavailable, usually mid-failover or briefly during a scale operation. Retry on a backoff; if it persists past a few minutes, capture the session tracing ID and open a support case.
42108	Can not connect to the SQL pool since it is paused. Please resume the SQL pool and try again.	The dedicated SQL pool (Synapse) is in a paused state. Retry succeeds only after the pool is resumed. Resume the pool explicitly, or schedule the workload to run after the pool resumes.
42109	The SQL pool is warming up. Please try again.	The dedicated SQL pool is resuming. Retry on a backoff until the pool is online; warmup typically takes a few minutes.
49918	Cannot process request. Not enough resources to process request. The service is currently busy. Please retry the request later.	The server can't currently allocate enough resources to satisfy the request. Retry on a backoff. If the error persists, scale up the database or elastic pool.
49919	Cannot process create or update request. Too many create or update operations in progress for subscription "%ld".	Subscription-level concurrency limit on management operations. Reduce parallel create/update calls or stagger them.
49920	Cannot process request. Too many operations in progress for subscription "%ld".	Subscription-level concurrency limit on operations in flight. Reduce parallelism or wait for in-flight operations to drain.

Statement-level errors aren't in this list because they fire after the connection is established and the failure leaves the session usable. The most common retryable statement errors are 1205 (deadlock victim) and 1222 (lock-request timeout). Retry the entire transaction rather than the single failing statement.

Error message text is from Azure SQL transient connection errors. Individual drivers maintain their own built-in retry lists; this catalog describes which errors are eligible for retry across SQL Server, Azure SQL Database, Azure SQL Managed Instance, SQL database in Microsoft Fabric, and dedicated SQL pools in Azure Synapse Analytics.