I have SQL server 2017. I have one 3 TB size database there. Somehow due to long running transaction the database got stuck 'IN Recovery' mode after SQL server Restarted. When I checked the sql error logs it says 2 189 255 seconds remaining (Phase 2 of 3) completed which is almost 25 days. My goal is to bring the database online even if I lose some data.

So I've ran below commands but no luck.

  USE [master]

  --Msg 3101, Level 16, State 1, Line 6
  --Exclusive access could not be obtained because the database is in use.
  --Msg 3013, Level 16, State 1, Line 6
   --RESTORE DATABASE is terminating abnormally.

  --Msg 5011, Level 14, State 7, Line 13
  --User does not have permission to alter database 'DragonDriveConnect', 
  the database does not exist, or the database is not in a state that 
  allows access checks.
  --Msg 5069, Level 16, State 1, Line 13
  --ALTER DATABASE statement failed.

  --Msg 922, Level 14, State 1, Line 22
  --Database 'DragonDriveConnect' is being recovered. Waiting until 
   recovery is finished.

At last I've tried to Delete the database too but that is also not working and giving me error saying Cant Delete.

How do I get out of this situation?


4 Answers 4


The error you're seeing in the SQL Server Error Log is this one:

Recovery of database 'CrashTestDummy' (9) is 0% complete (approximately 42 seconds remain). Phase 2 of 3. This is an informational message only. No user action is required

More generically, it will say:

Recovery of database '{Database Name}' ({Database ID}) is {N}% complete (approximately {N} seconds remain). Phase {N} of 3. This is an informational message only. No user action is required

Because your database was not shut down cleanly on restart, the database must go through "crash recovery". This is required to ensure the database remains consistent. When the database is not shut down cleanly, SQL Server must make sure that transactions written to the transaction log have been reconciled properly against the data files.

All transactions are guaranteed to be written to the transaction log. However, updating data is initially only done in memory. Updates to the physical data files are done asynchronously via a checkpoint. The asynchronous nature of the data file updates are why a crash or unclean shutdown requires extra work on startup.

As indicated by the error message, there are three phases to the recovery. Each of these is essentially a pass through the transaction log:

  1. Analysis
  2. Redo / Roll Forward
  3. Undo / Rollback


This phase is simply to review the transaction log & determine what needs to be done. It will identify when the most recent checkpoint was, and what transactions might need to be rolled forward or back to ensure consistency.

Redo / Roll Forward

Completed transactions from the transaction log need to be reviewed to ensure the data file has been completed updated. Without this, changes that were only in-memory might have been lost.

This phase will take those transactions that were committed after the most recent checkpoint and redo them, to ensure they are persisted to the data file.

If you are using SQL Server Enterprise edition, Fast Recovery will allow the database to come online & be available after this phase of recovery. If you are not using Enterprise Edition, the database will not be available until after the Undo phase has completed.

Undo / Rollback

Transactions from the transaction log that were rolled back, or were uncommitted at the time of "crash" must be rolled back. SQL Server must verify that if uncommitted changes were made to the data file, they are undone. Without this, a rolled back change could be partially committed, violating the ACID principles of the database.

This phase will perform the rollback of any transactions that were uncommitted at the time of crash, or were rolled back after the final checkpoint.

So what can you do about it?

While the database is in recovery, attempts to bring the database online via a RESTORE command like this will fail:


SQL Server is already attempting to do this. The RESTORE...WITH RECOVERY; will simply put the database through the exact same steps in order to bring the database online in a consistent manner.

Be patient

The right thing to do is just be patient. This part of the message from the error log is the one that you should pay attention to:

No user action is required

Note, too, that the time remaining is an estimate. In my experience, it is wildly inaccurate. Sometimes the time remaining will grow larger, rather than reducing. Sometimes it will report a very long completion time, and suddenly complete very fast. Its just an estimate.

Can you just "throw away" your transaction log and start fresh?

I advise against it. I would suggest you never, ever do this with a production database. There is a procedure to attach a database without a transaction log, and ask SQL Server to ATTACH_REBUILD_LOG. I won't detail all the steps, but the "punchline" for that procedure is to do this:


Running this on a crashed database may result in this error:

The log cannot be rebuilt because there were open transactions/users when the database was shutdown, no checkpoint occurred to the database, or the database was read-only. This error could occur if the transaction log file was manually deleted or lost due to a hardware or environment failure. Msg 1813, Level 16, State 2, Line 5 Could not open new database 'CrashTestDummy'. CREATE DATABASE is aborted.

In which case, you're stuck. You'll need to use the original transaction log & be patient. Just wait for it to recover.

  • Thanks a lot. I stopped the server. Remove mdf and ldf files and was able to delete the database. Nothing else worked in my case. Thanks for the detailed analysis and explaination though.
    – nicktheman
    Commented Oct 13, 2019 at 1:32
  • In my (weird) case, there was no sign of recovery progress in logs. Restarting of the engine did not help. The problem resolved completely after restarting the OS.
    – Ahmad
    Commented Sep 28, 2023 at 19:04

Patience, patience!

Possible causes for the databases to stuck in “In Recovery” mode?

  • Huge size of transaction log file.
  • SQL restarted during a long running transaction.
  • Huge number of VLFs (i.e. virtual Log Files).
  • Could be a bug in SQL Server which is fixed with the help of some patches.

The very first thing that your should do is checking of ERRORLOG. In Errorlog, we should see the very first message in the database (TestMe is the name of my database):

Starting up database ‘TestMe’.

This means the files are opened and recovery is started. After sometime, you should see phase 1.

Recovery of database ‘TestMe’ (28) is 0% complete (approximately 37 seconds remain). Phase 1 of 3. This is an informational message only. No user action is required.
Recovery of database ‘TestMe’ (28) is 3% complete (approximately 36 seconds remain). Phase 1 of 3. This is an informational message only. No user action is required.

Once phase 1 is complete, it would go with Phase 2 and 3.

And once it completes, you should use something similar.

3807 transactions rolled forward in database ‘TestMe’ (28). This is an informational message only. No user action is required.
0 transactions rolled back in database ‘TestMe’ (28). This is an informational message only. No user action is required.
Recovery is writing a checkpoint in database ‘TestMe’ (28). This is an informational message only. No user action is required.
Recovery completed for database TestMe (database ID 28) in 30 second(s) (analysis 1289 ms, redo 29343 ms, undo 72 ms.) This is an informational message only. No user action is required
  • Thanks but if you read my question, I've done all that and in the ERRORLOG, it says recovery will complete after 25 days. Its not acceptable at any cost.
    – nicktheman
    Commented Oct 14, 2019 at 18:22

If you have SQL Server 2019 you can enable accelerated database recovery which is many times faster and can help avoid this problem in the future


  • Please try and add some (of the) relevant information from the link to your answer. If you remove the link, your answer becomes a simple comment with no added value. Links can become obsolete. Thanks.
    – John K. N.
    Commented Sep 10, 2021 at 6:30
  • 1
    John I see what you're trying to say but isn't the relevant information (MVP) is that a new feature exists, which version it was introduced in and what it is called. Although Microsoft might, sorry will eventually break the link the feature will still be documented online and indexed by the search engines Commented Sep 10, 2021 at 7:18

In my case, This issue has been resolved by Restarting SQL Server Service, You can do it from SQL Server Configuration Manager.

enter image description here

  • Glad this worked for you. If the service is actually busy recovering databases restarting the service will only delay this work. If for you it was simply a "software's not refreshing" issue I can see how restarting may help. It's difficult to tell the difference without deeper analysis, though. Commented Sep 10, 2021 at 5:22
  • In my case this happened to help as well -- I had a RESTORE ... WITH STANDBY that was taking a preposterously long time. Recycling the server basically put it back into NORECOVERY and try again / restore some additional log backups. (I'm was not able to figure out exactly what the issue was that caused the pathological WITH STANDBY, but this helped move past it)
    – Mark Sowul
    Commented Jan 31 at 22:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.