How to fix Error 3421 on Windows Server 2025 / Windows 11

by

Troubleshooting Guide: SQL Server Error 3421 — Recovery Completed Message with Anomalies

Why This Error Happens

This message is logged by the SQL Server Engine during the database recovery process, which runs automatically whenever a database is brought online after a shutdown, failover, or restart. While often informational, when paired with unusually long analysis or redo times, it typically indicates that the Transaction Log (LDF) contains a large number of uncommitted transactions that need to be rolled back, or that the underlying storage subsystem is experiencing I/O latency. In more severe cases, this delay is a precursor to log corruption caused by an unexpected power loss or a faulty storage driver on Windows Server 2025.

Step-by-Step Resolution Guide

Step 1: Analyze the SQL Server Error Log

  • Open SQL Server Management Studio (SSMS) and navigate to Management > SQL Server Logs.
  • Look for the specific phases: “Analysis,” “Redo,” and “Undo” to identify which specific phase is consuming the most time (I64d ms).
  • Cross-reference the timestamp with Windows Event Viewer (Event ID 3421) to check for concurrent disk warnings.

Step 2: Check Storage Subsystem Health and Latency

  • Open Resource Monitor on Windows Server 2025 and check the “Disk” tab for high queue lengths on the drive hosting your .mdf and .ldf files.
  • Verify that the storage drivers are up to date via Device Manager > Disk Drives.
  • Run the following T-SQL query to check for I/O stalls:
SELECT * FROM sys.dm_io_virtual_file_stats(DB_ID('YourDatabaseName'), NULL);

Step 3: Increase Recovery Interval Configuration

  • If the database is large and checkpoints aren’t frequent enough, manually adjust the recovery interval to force more frequent checkpoints, reducing future recovery times.
  • Execute the following command in a New Query window:
EXEC sp_configure 'recovery interval', '0';
RECONFIGURE;
  • Restart the SQL Server service via services.msc for the changes to take effect fully.

Step 4: Validate Database Integrity

  • Run a full consistency check to ensure the lengthy recovery didn’t mask underlying corruption in the data pages.
  • Use the following command:
DBCC CHECKDB ('YourDatabaseName') WITH NO_INFOMSGS, ALL_ERRORMSGS;
  • If errors are returned, proceed to restore from the last known good backup rather than using repair options that might cause data loss.

Alternative or Verification Steps

  • Verify the Fix: Restart the SQL Server service (MSSQLSERVER) via PowerShell using Restart-Service MSSQLSERVER and monitor the Error Log to confirm the “Recovery completed” time has significantly decreased.
  • Alternative Solution: If recovery times remain high due to hardware limitations, consider migrating the transaction log file to a dedicated high-speed SSD volume separate from the data files to reduce I/O contention.