Microsoft Exchange 2000 Operations Guide — Version 1.0
84
Once again, you would be advised to define your SLAs so that mailbox recovery is not
possible outside the period of time you specified in the Administrator program. While
mailbox recovery is possible outside of this time span, dependent upon your backup
software, you may have to restore an entire Exchange database to a server in a different
Windows 2000 forest to get the appropriate missing mailbox.
Recovering Exchange Stores and Storage Groups
In this scenario, there has been a problem with the database, for example a corruption in
one of the databases and you need to restore them from backup. The rest of your environ-
ment has not been affected.
Before doing anything else here, if you are recovering from tape, you should ensure that
you have made copies of your existing database files. It may be that when you have
recovered from tape, you discover that the tape is bad. You may be able to recover the files
that you archived by using troubleshooting techniques, even though the database has a
problem. If you always make sure that you never let your database drive become more
than half full, then you can quickly save a copy of a database that “crashes” on the same
logical drive, dramatically decreasing the time it takes to copy the files, and therefore your
recovery time.
When you come to do the restore, you will need to ensure that the information store
service is started, and the databases you want to restore are dismounted. You will need to
select a temporary folder for the restore. This will contain restored log and patch files,
alongside restore.env, a binary file which ensures that the log and patch files are replayed
properly at the end of the restore.
Assuming you are recovering a full backup (as opposed to incremental or differential) you
should ensure that you select the Last Restore Set in the backup set. This ensures that the
log files and patch files are replayed after the restore, taking you more or less to the point
of failure. You will not be able to mount the databases unless the last restore set option is
checked, so if you forget to do this, you will either have to run the restore again, or use
Eseutil to specify that it is the last restore set.
If you are restoring multiple storage groups simultaneously, you must specify a different
temporary folder for each one. This ensures that the different restore.envs do not overwrite
one another.
Restoring an offline backup is not generally recommended because it does not allow you to
roll forward to the current status. However, it can be useful in a situation where a restore
from an online backup has failed. The important thing to realize here is that the .edb and
.stm files should be regarded as one and restored together in the same directory. Also,
when performing the restore, you should delete all log and database files on the recovery
server before copying over the new ones. The recovery server will create its own new log
files when you start the services up again.
