Using Exmerge to recreate priv1.edb in Exchange 2000

I skipped Exch2000, and am using 2003. But I don't think ExMerge has changed much.
Check this article on importing:
http://www.dtidata.com/resourcecenter/2007/05/11/exchange-server-tools-use-exmerge-import-pst-files/

It is a long and painful process, but it can be done.
It will also take a lot of disk space.

See this page for step-by-step instructions.
http://support.microsoft.com/kb/265441

I have to spend some time with the article, but I guess the part I'm confused about is how to create a clean database with my existing users without reentering all the info, and similarly, how to keep the transaction logs in place and not lose them.

Mp022, from reading this it seems that my pub.edb files and transaction logs stay in place?  Is that true?

Also, the article says "start the information store to create a new priv.edb".  How exactly do I do that, and what do I do with the old priv1.edb and .stm files beforehand?

what version of exhcange do you have standard or enterprise
if you have enterprise you can create a new database and move the mailboxes

you would need to create the new database with the identical name to avoid issues if you went the export and import

Standard.  Unfortunately.  I'm a home user with 8 mailboxes and a 12G priv1 file.

then yes, use exmerge to get everything to PST
I wouldn't bother deleting the database from the console
i would just delete the database and log files
when you attempt to mount the database it will warn you that it is mounting a blank database
then you can import

I replaced some hardware (I was gettting event 50 errors) and am now running eseutil /p (painfully slowly), so let me try that first before I resort to an export import strategy.  By way of background, what led me to discover the corruption in the database was that ntbackup aborted with checksum errors (error 1018) (I had previously just been backing up the raw edb and stm files).

In your opinion, what do I do assuming the repair is "successful", meaning eseutil /p finishes without error:

1.  Try and mount the priv1.edb without backing it up first and see what happens;
2.  Try and backup the priv1.edb and log files by trying to see if I can back up the entire "information store" in ntbackup
3.  Just run eseutil /d on the repaired priv1.edb file?
4.  Something else?

Thanks for your help.  I am obviously not an IT pro!!!

that is normally a hardware issue (1018)
i would have replaced the hardware and attempted a restore, always restore before a repair
you may want to restore after the /p

1. you must run an eseutil /d to defrag the database, then run an isinteg
2. yes backup after everything is done

you want to move to a new database after running /p (best practice if possible)

I have no idea how to do that, so I want to take one step at a time.

1.  The database would mount - it just wouldn't backup using ntbackup with "information store" checked because of checksum errors;
2.   I ran eseutil /k and it said 4 pages had checksum errors;
3.   I ran eseutil /d and it aborted because of checksum errors;
4.   I am now (after a number of false starts due to error 50's) runnding eseutil /p successfuly (hopefullly, I'm not jinxing anything)l
5.   Next you suggest I run ntbackup of the entire information store BEFORE I try and mount the repaired (hopefully) epriv1.edb file and defrag it, or after?

Sorry, I'm just trying to understand...

as long as you had a backup and all of the logs since the backup, you could attempt to restore with the new hardware in place

you need to defrag and run the isinteg before mounting the database

Just to be clear, I had no ntback of the "information store" to restore because I  (stupidly) backed up only the 4 .ebd and .stm files.  All my backups had the same "checksum" 1018 error in them.  It was only when I realized that I was not backing up proprerly that I discovered the need to repair the database file.  Am I clear?

That is exactly my error in backup - I never backed up any logs - just the .edb and .stm files that I store on a USB drive connected to the server.  How do I do that (backup the logs)?  And by "restore" do you mean to mount the repaired data base?  Sorry to be so dense!

np, you are on the right path for your situation
finish the repair with the /p, run the defrag and isinteg
you should be able to mount the database at that point

http://support.microsoft.com/kb/259851

So no intermediate backups at all?

you can pick a backup strategy that best suits your needs
the most popular is a full backup one day with incremental backups the other days

Sorry, I am not asking about a general backup philosophy.  I am asking about whether I need to back up at this point in the recovery before I attempt to mount the repaired database, and if so, how?

Thanks.  Will it hurt to add an additional ntbackup between steps 1 and 2 so I have a backup of the current state of the transaction logs, or is that a waste of time?

waste of time, look at the article http://support.microsoft.com/kb/259851
you should run the defrag and integrity check before doing anything else with the db

I took a look at the priv1.integ.raw file and all /p did was delete 4 pages with bad checksums.  I think it also rebuilt 2 or 3 indexes.  No errors in the stm file at all.  Given that is it really necessary to run isinteg after defrag finishes (each proceedure seems to take about 15 hours to complete!)?

yes, how big is the database? the defrag shouldn't be as long as the /p

12G.  I wasn't asking about the defrag.  I was actually asking about the ISInteg -fix

yes, you need to run each step. it is best practice when repairing a database.

Ok.  Got it.  Thanks Endital1097.

One more related question if you don't mind:  When I run NTBACKUP to backup the mailstore, what happens (assuming I overwrite the media) if the backup is interrupted.  Does it first backup to a temp file and then copy that over to overwrite the last successful backup, or does it destroy the first backup before it writes a new one?

Basically, I'm wondering if I do a full backup every day whether I need to rotate backups with one running every other day, and the other running on the days inbetween, so as to always have a spare set if the backup is destroyed before a new one is successfully written.

you lose what is on the media with an overwrite, so you should have at least two backup locations (tapes/drives)

I lose it BEFORE the new backup completes?

yes, it must make room for the new data

Is there some kind of batch file I can add as a scheduled task that will copy the old backup to a different location after it is created and before the new backup is started the next day?  If so, how would I do that?

BTW, thanks so much for all your help!!

what is your backup media?
you could always have a scheduled task for each day of the week that uses a different backup location (or every other day does)

It's an NAS on the LAN.

I'm thinking maybe I should just do a full backup manually monthly (?) and then set a task to do a differential backup each night?

What do you think of that as opposed to one full backup job that does M, W, F and another that does T,Th, and Sat?

So only full backups trancate the logs?

And if I restore from a differential, it will replay all logs since the last full backup?

an incremental backup does as well, just not a copy or differential

yes, it must play all logs from all restores required to get the database up-to-date

a full backup would have the least of course

If I have disk space, what do you think about doing a full backup each night and appending to the existing media rather than overwriting it?

I am just worried if I go away for a week or two, and have two rotating full backups on alternate days, and the backup gets corrupted somehow, I'll have no restorable backup by the  time I get back and am able to check.

Or is there a way to get NTBACKUP to keep creating full backups but to number them sequentially (like differentials).  For Example, the original backup would be Mailstore1.bkf, the next night it would create Mailstore2.bkf, etc?

i would use two and write a full plus incrementals

but ultimately you should do what you feel most comfortable with

you would need to create separate backup jobs to have the different file names

Sorry, I don't understand this:

"i would use two and write a full plus incrementals"

Could  you please explain?

Ok, I'm all set.  Datastore is restored and backup up!  Thank you so much!!

Endital1097 was VERY helpful and timely in his responses!

Great news. Enjoy the rest of your weekend