[Solved] Full NTbackup of Exchange 2003 restores but fails to remount on last week's SBS2003 disk image

Question

Edit - solved - needed to remove present mdbdata folder contents before starting a restore.
Bit of a long post, sorry - trying to tell the whole story.
I run SBS2003 in a virtual machine. I stop the VM and copy the virtual disk file to another drive once a week as part of the backup routine. I also take full ("Normal") online NTbackup of Exchange each night. There is only one storage group, containing
a mailbox store and a public folder store. At one time I tested the restore process for exchange and it worked as I recollect, but it isn't now.
I took one of my backups from last week and restored last night's full Exchange backup from the active server. I dismounted the stores in Exchange System Manager (ESM), marked the databases as able to be replaced by a backup, and ran the bkf file in NTbackup.
I was able to watch the files being replaced and rebuilding in the explorer window while the NTbackup was running. I had marked the NTbackup as "Last Restore" and "Mount after Restoring". The NTbackup restore process completes successfully.
When I go to ESM I find that the stores are not mounted. I try to mount them manually and get an error c1041724 from ESM on each store, saying restart ESM or the Information Store Service. Restarting both still gives me the c1041724 in ESM when I try to mount
the stores. restarting the server vm completely leaves me with the stores still unmounted, same errors from ESM.
I check the application events after I try to mount the stores, and I get these events:
1. ESE event 100 database engine started.
2. ESE event 102 database engine started a new instance.
3. ESE event 619 database from backup set on which hard recovery was not started or did not complete successfully
4. MSExchangeS event 9519 error 0xfffffde0 starting database
5. MSExchangeS event 9518 error 0xfffffde0 starting Storage Group.....MDB failed to start
then two ESE events saying database engine stopped.
RE event 619, http://support.microsoft.com/kb/810199, case 1, says "This issue may occur if you do not select the Last Backup Set check box."
http://support.microsoft.com/kb/253931 says the same thing, Last Backup Set not checked during the restore. But I did select that box. (FYI, NTbackup calls it "Last Restore Set" - same thing?)
RE event 9519 error 0xfffffde0,
http://support.microsoft.com/kb/258529 says "the Storage Group attempted to restart after a crash, and that the database was restored (but not hard-recovered). In other words, the hard recovery log file replay process was not successfully completed,
or was possibly interrupted."
RE event 619, http://support.microsoft.com/kb/810199, case 1, says to run eseutil /cc on the temporary path where restore.env is located. I do so and get error -939523485. I am directed to
http://support.microsoft.com/kb/328639/EN-US and I see the four events on the KB article are in my event viewer. The article says "old log files are out of sequence with the log that is mentioned
in Restore.env" Event 458 says "The logfiles E0005C75.log and E0005C9D.log are not in a valid sequence". Log file E0005C75.log is in the temporary folder with the Restore.env file. There is no E0005C9D.log in either the temporary folder or the
mdbdata folder. The mdbdata folder has only an E00.log and an E00.chk, and res1.log and res2.log. I do find in the working server that there is a E0005C9D.log.
It sounds like I'm missing some logs from the backup. But it's supposed to be a full Normal backup. I find no mention anywhere that one needs to go back into the running server to copy logs out. What am I missing?

Answer 1

OK, so here's something interesting. Noting that the errors implied I was missing logs (from my "full" backup-?) I said why not also back up the whole exchange database folder too? Then I might get those missing logs. So I set NTbackup
to do an online backup of the Information Store, and the exchange mdbdata folder's contents, in the same bkf.
I try to restore the bkf first to my test server, doing just the information store first and leaving the extra mdbdata folder unchecked. Also, the "Last Restore" box is UNchecked so the logs won't be replayed when the restore is done. After that
restore runs, I then uncheck the information store backup and check the mdbdata folder, and run the restore again. The new logs come in now. (I can't run them both in the same restore, information store has to be first, then mdbdata folder.)
After NTbackup is finished with both restores I go to ESM, right-click the stores and Mount them - no eseutil commands. The databases mount just fine, no errors, and I can log in with remote exchange on a test client's IE, no issues; all emails and
calendar posts from the previous day, up to 5PM when the backup runs, are present. The only problem, besides no Microsoft or other websites that I can find saying a full backup needs a supplemental backup to be full, is that my backups are twice
as big as they were before, from copying the databases twice.
So a "full" backup from NTbackup isn't full enough, and a non-documented shot in the dark gets the restore to work.
Is my server issue-matic, or is something not right here with NTbackup?