SQL Database Repair Guide: Fix Corrupt MDF Files [2026]

Your database goes into suspect mode. Panic sets in. The error messages on your screen seem to be speaking a different language, and the clock on downtime is ticking loudly. You’ve lost access to critical data, and if you’re like most DBAs I’ve spoken to over the last decade, the first thought isn’t a command—you’re wondering if this is the end of the line.

Here is exactly how to fix it without losing data.

When facing a corrupt SQL database fix, there is no magic button, but there is a hierarchy of solutions. From diagnosing the specific I/O errors to running internal repair commands, and finally to using specialized third-party tools, the path to recovery is clear if you stay calm and systematic. While this guide primarily focuses on the Microsoft SQL Server ecosystem where MDF/NDF corruption is most common, the principles of sql database repair often overlap with MySQL contexts when dealing with InnoDB file groups. Let’s walk through the process step-by-step.

Diagnosing the Damage: Identifying a Corrupt SQL Database

Before you can repair a database, you need to know exactly what you’re fighting. Is it a bad page? A broken index? Or is the transaction log irretrievably damaged? In my experience, jumping straight to repair without a solid diagnosis is like performing surgery without an X-ray.

Common Error Codes and Symptoms

SQL Server is notoriously verbose when things go wrong. Learning to read these errors is half the battle. When you see Error 824 or 825, you are dealing with an I/O subsystem failure. This means the physical disk hardware or the operating system is failing to read/write data correctly. It’s not just the database engine complaining; it’s the floor beneath it shaking.

Error 9004 is another frequent offender, typically indicating that the database is in an inconsistent state due to these I/O errors.

Then there is the dreaded Suspect Mode. When a database enters this state, SQL Server cannot complete the startup recovery process. This usually happens because of severe corruption, a missing file, or a failed shutdown that left transactions uncommitted. You might also see Recovery Pending, which suggests the server couldn't start recovery due to lack of resources, such as disk space.

Another specific error to watch for is Error 5171. This message—"MDF is not a primary file"—usually appears when you try to attach a secondary data file (.ndf) or a copy of a data file that doesn't contain the system table headers. It’s a common mistake when people try to manually re-attach files from a disorganized backup folder.

Running DBCC CHECKDB for Integrity Verification

Once you’ve identified the symptoms, DBCC CHECKDB is your primary diagnostic tool. It checks both the physical and logical consistency of the database. Think of it as a full-body scan for your data.

To run it, you simply execute the command against your corrupt database. However, be prepared—it can take a long time on large databases. Here is the basic syntax:

USE master;
GO
DBCC CHECKDB ('YourCorruptDatabaseName') 
WITH NO_INFOMSGS, ALL_ERRORMSGS;
GO

The output will break down into two main categories: allocation errors and consistency errors. Allocation errors usually point to physical corruption—pages that are marked as allocated but don’t exist, or pages that contain data but aren’t referenced by any index. Consistency errors are logical issues, such as a foreign key constraint pointing to a non-existent row or a mismatched row count.

Distinguishing between physical corruption (bad sectors on disk, torn pages) and logical corruption (broken pointers, index mismatches) is crucial. Physical corruption is harder to fix because the underlying bits are wrong. Logical corruption, while scary, is sometimes easier for repair tools to reconstruct because the data might still be intact, just poorly organized.

In one of my earlier cases, a client’s database was stuck in suspect mode. A quick CHECKDB revealed only logical indexing errors, not physical page damage. We were able to repair it with minimal data loss. Had we assumed the worst, we might have gone straight to expensive third-party tools.

How to Fix a Corrupt SQL Database Without Backup

Let’s address the elephant in the room: you don’t have a recent backup. Maybe the backups were never tested, or the storage failed along with the primary database. In this scenario, your corrupt sql database fix options become more limited and riskier.

Repairing MDF Files via ALTER DATABASE

If DBCC CHECKDB confirms corruption but you have no clean backup, you can attempt to repair the MDF file directly. This is a last-resort measure and should be taken with extreme caution.

First, you need to isolate the database. Run the following commands to set the database to SINGLE_USER mode. This prevents other users from connecting and potentially making things worse.

USE master;
GO
ALTER DATABASE YourCorruptDatabaseName 
SET SINGLE_USER 
WITH ROLLBACK IMMEDIATE;
GO

Next, you run the repair command. There are two main options: REPAIR_REBUILD and REPAIR_ALLOW_DATA_LOSS. The former fixes minor issues like missing index rows without losing data. The latter is the nuclear option. It will fix allocation errors and row/link mismatches, but it may delete corrupted data to restore structural integrity.

DBCC CHECKDB ('YourCorruptDatabaseName', REPAIR_ALLOW_DATA_LOSS) 
WITH ALL_ERRORMSGS;
GO

Warning: This command can result in permanent data loss. It is repairing the database structure, not the data itself. If a page is physically unreadable, REPAIR_ALLOW_DATA_LOSS will likely drop the affected rows. Always try REPAIR_REBUILD first if CHECKDB allows it. Once the repair is complete, remember to set the database back to MULTI_USER mode.

Restoring from Suspect Mode

Sometimes, the database is simply stuck in a loop during recovery. You can force it into EMERGENCY mode to make it readable for data extraction. This puts the database into a read-only state, allowing you to script out your tables or copy data to a new database.

ALTER DATABASE YourCorruptDatabaseName SET EMERGENCY;
GO
ALTER DATABASE YourCorruptDatabaseName SET SINGLE_USER;
GO
DBCC CHECKDB ('YourCorruptDatabaseName', REPAIR_ALLOW_DATA_LOSS)
GO
ALTER DATABASE YourCorruptDatabaseName SET MULTI_USER;
GO

This checklist is critical when dealing with suspect mode:

1. Backup the .mdf and .ldf files immediately. Even if they are corrupt, having a raw copy is essential for third-party tool recovery.
2. Set to EMERGENCY mode to bypass normal recovery checks.
3. Run DBCC CHECKDB to understand the extent of the damage.
4. If repair fails, detach the database and attempt to use a specialized recovery tool on the raw files.

I’ve seen many DBAs try to force-attach a detached MDF file without checking the log file first. If the transaction log is mismatched, SQL Server will reject the attachment. In those cases, you often need to create a new empty database and use import/export wizards or third-party tools to move the readable data out.

Top SQL Database Recovery Tools for Windows (2026 Review)

When built-in commands fail, you need specialized sql database recovery tools. The market is flooded with options, but not all are created equal. In 2026, the landscape has shifted slightly towards tools that offer better preview capabilities and support for both SQL Server and MySQL environments.

Comparing Free vs. Paid Solutions

Built-in Microsoft tools are free, but they have significant limitations. They require you to be able to attach the database, which is impossible if the metadata is too corrupted. They also don’t offer a preview, meaning you might spend hours running a repair only to find out the data is still inaccessible.

Third-party tools like Stellar Repair for MS SQL, SysTools SQL Recovery, and ACE Data Recovery’s software suite offer more robust engines. They can often parse MDF files directly without needing to attach them to a running SQL Server instance.

Here is a quick comparison:

Best Tools for SQL Server and MySQL

One thing many competitors miss is that sql database repair isn’t just about SQL Server. Many organizations use MySQL for their web applications, and corruption in .ibd files follows similar patterns.

Tools like ACE DRS and some versions of Stellar now support both formats. When choosing a tool, look for:

• Deep Scan Capability: Can it reconstruct data from raw bytes if the header is gone?
• Preview Before Save: This is non-negotiable. You need to verify your data is intact before paying for the full version.
• Filegroup Support: Does it handle complex multi-filegroup databases?

In my testing, tools that allow you to save the recovered data directly to a new SQL Server instance or as a .bak file are the most useful. They streamline the restoration process, getting your application back online faster.

Preventive Measures: Ensuring Database Integrity Going Forward

The best sql database repair is the one you never have to do. Prevention is cheaper than cure, both in terms of money and sleepless nights.

Automated Maintenance and Backup Strategies

You should be running a database integrity check on a regular schedule. I recommend a weekly DBCC CHECKDB on all user databases. For critical production systems, consider daily checks. Automate this using SQL Agent jobs so you get an email report if anything fails.

Backup strategies are equally important. Understand the difference between full backups and transaction log backups. A full backup captures the state of the database at a specific point in time. Transaction log backups capture every change made since the last backup. By combining these, you can achieve point-in-time recovery.

Store your .BAK files off-site or in cloud storage. If your server room floods, having backups on the same rack is useless. Use tools like Azure Blob Storage or AWS S3 for durable, inexpensive off-site storage.

Understanding Transaction Logs and Filegroups

The transaction log is your safety net. It allows you to recover to a specific second before a crash. Ensure your log backups are running frequently. If your log file is full, your database might refuse writes or enter recovery pending states.

Filegroups can also improve resilience. By placing indexes on one filegroup and data on another, you can isolate corruption. If an index filegroup gets corrupted, you might still be able to access the data in the primary filegroup. This separation of concerns is a best practice for large, enterprise-level databases.

FAQ

Can you repair an MDF file without a backup? Yes, but with significant caveats. You can use the DBCC CHECKDB command with the REPAIR_ALLOW_DATA_LOSS option. However, this is a last resort and may result in the loss of corrupted data. For better results without a backup, specialized third-party SQL repair tools can often recover data from corrupt MDF files by scanning the raw file structure.

How do I fix a corrupted SQL database in suspect mode? To fix a database in suspect mode, you first need to bring it to a manageable state. Run the following commands:

1. ALTER DATABASE [YourDB] SET EMERGENCY;
2. ALTER DATABASE [YourDB] SET SINGLE_USER;
3. DBCC CHECKDB ([YourDB], REPAIR_ALLOW_DATA_LOSS);
4. ALTER DATABASE [YourDB] SET MULTI_USER; This process attempts to repair the database and make it accessible again.

What is the difference between database restore and repair? Restore replaces your current damaged database files with a known good copy from a backup. It is the preferred method because it ensures data integrity. Repair attempts to fix the current damaged files in place. Repair is used when no backup is available, but it carries a higher risk of data loss and may not succeed if the corruption is severe.

How much does it cost to recover a corrupted SQL database? Costs vary widely. DIY methods using built-in SQL commands are free. Commercial sql database recovery tools typically range from $200 to $500 per license. Professional data recovery services, especially for physically damaged drives, can cost anywhere from $1,000 to $5,000 or more, depending on the complexity and urgency of the recovery.

Conclusion

Database corruption is a nightmare scenario, but it’s not always fatal. By understanding how to diagnose the damage, utilizing built-in commands like DBCC CHECKDB, and knowing when to reach for professional sql database recovery tools, you can maximize your chances of saving your data.

While manual repairs are useful for minor issues, severe corruption often requires the advanced parsing capabilities of dedicated software. Remember, the hierarchy of solutions matters: diagnose first, then attempt repair, and finally resort to third-party tools if necessary. Most importantly, invest in prevention. A robust database integrity check schedule and reliable backup strategy are your best defenses against future outages.

If you are currently facing a corruption crisis, I recommend downloading a free trial of a reputable repair tool to assess the damage before committing to a purchase. For long-term peace of mind, consider implementing automated maintenance plans today.