Restore using NTBACKUP (Exchange)
This page details the process of restoring a Microsoft Exchange database using NTBACKUP and replaying the log files back into the database whilst overcoming some common errors. The guide below already presumes the user has some knowledge of Exchange and that the replacement server has been rebuilt, and Exchange installed. The server must have the same service pack update as the server where the database was originally hosted.
- Rename the mailbox store to match the name of the mailbox store on server whose database is being restored - even if it includes the hostname of the previous server. NTBackup identifies which store to restore to based on this string.
- From Exchange System Manager, open properties of the information store(s) where a restore is being attempted and tick 'this database can be overwritten by a restore.'
- In NTBackup, catalogue the BKF file and highlight Logs and mailbox store as a minimum. Only restore the public folder store if the previous server is dead or will never be booted whilst connected to the same forest/domain; otherwise public folder and GAL replication issues among others are likely to occur. Choose an empty temporary directory to work from and remember it.
- Choose the appropriate destination server but do not choose log file replay even if this is the last backup set - it is better to do it manually as there are no verbose-logging of it if it fails when it is performed by NTBackup.
- When the backup has finished check the temporary directory for a file - restore.env. If it is not present the restore will need to be retried as it is vital to restore the Exchange DB to the point just before it failed. - Retry the restore in the first instance and if an error occurs that prevents a decent restore.env file from being created or none at all see When Log file replay/hard recovery is not available.
- Copy all the log files available as they are without renaming the files to the same directory as the IS database files.
- Cd to the eseutil folder, c:\program files\exchsvr\bin by default; and type eseutil /cc "path to temporary dir containing restore.env" This replays the transactions logs back into the database without checking the checkpoint file as it is likely to have been lost anyhow. If you receive a JET errBadLogVersion error see: Error -514 JET errBadLogVersion when using Eseutil
- The process normally takes minutes and if the readout states that there are no errors go on to the next step; else look at When Log file replay/hard recovery is not available.
- From Exchange System Manager go to the mailbox store and the mailboxes should all be seen. Right click a blank area and choose Run Clean-up agent. All the mailboxes should now have a red-cross on them as they are currently not associated/connected to a user account in AD. An Exchange mailbox must be connected to a user account before it can be opened. Right-click the mailbox and choose reconnect. If the following error appears: "The operation cannot be performed because this mailbox was already connected to another user ID no: c1034ad6" see Cannot reconnect mailboxes.
- Once the mailbox is reconnected it can be opened as usual.
- Exchange 2003 Backup and Restore with NTBACKUP
- Exchange 2007 restore using NTBackup
- NTBACKUP caveat for Exchange and system state backups
- Backup Operation Fails When You Back Up Exchange Server 2003 Databases and System State Information at the Same Time
When Log file replay/hard recovery is not available
Depending on how log file replay has failed the simplest possible method to allow it to work is to truncate the logs:
- Initially delete E00.tmp if it exists and try again - this is the name of the log that Exchange is/was currently writing to and is likley to be corrupt in this scenario.
- Delete E00.log and try again - this was the last log that Exchange finished writing to.
- Delete the next 10 most recent logs and try again. In some scenarios it is required to rename the latest log file to E00.log .
If this still fails try the following:
Eseutil /ml <log file> verifies a log file and can give info that can be googled that when applied could allow a log file replay to partially work. eseutil /k <log file> as above but checks headers only
Repairing log files
Log file repair should only be attempted if all of the above has been tried. If a repair of the log files is attempted it can make the scenario worse. - To save having to do a re-restore perform an offline backup - copy the files elsewhere. Never try this on a live server without performing an offline backup. If existing transaction log files are not found by Eseutil when it tries to run recovery or they are cannot be recovered, it will create a new transaction log file, and try to attach the database to it. If the database is Inconsistent or in Dirty Shutdown state, the database will not be made startable. If the database is in a consistent state, it will be attached and then detached from the new log file. - In either of these cases you risk making changes to the database or adding log files to the server that will result in the database becoming unstartable or that will confuse further recovery troubleshooting.
Type: eseutil /r E00 /i
If log file reapir states it was a success this does not necessarily mean there was a success. Recovery succeeds whenever all available transaction log data that is currently available has been applied to the database files. Recovery success says nothing about whether the available data was sufficient to restore the databases to consistency.
Repairing the database
The last attempt to enable log file replay to work is to perform a /p on the database - repair mode. DBs who have a /p applied should never be placed into production for an extended period of time. Normally a /p gives no more then restoring a DB and not attempting log file replay however it can allow log file replay to work - but only in a very tentative state. Always move mailboxes to another store even if this works.
Type: eseutil /p priv1.edb
If a streaming miss-match occurs try again with the /i switch also.
After this you must run eseutil /d to defrag the DB followed by isinteg -fix . Only if you intend to use the repaired database for salvage only can you skip these extra steps. Skipping them means that you may salvage less data than if you completed them, but it can also mean saving several hours of recovery time.
Cannot reconnect mailboxes
When reconnecting mailboxes from a recently restored database the orphaned mailbox can still be associated with the original user account even though it appears to be disconnected. To resolve this:
Use ADSI Edit to change the
legacyExchangeDN attribute for the active user for the mailbox you wish to reconnect as System manager look for users with this attribute before reconnecting. When disconnecting a mailbox from a user this attribute would normally cleared and would not be an issue. I would recommend adding a suffix to the end of the attribute rather than removing it. Normally, changing this attribute would cause email for the user to stop working until reverted.
If this fails to work remove the Exchange attributes for the user: Right click the user in AD, choose Exchange tasks followed by Remove Exchange attributes. The mailbox can now be reconnected.