• Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 326
  • Last Modified:

Exchange 5.5 crash recovery with only dir.edb available (how to set up an emtpy system with just dir.edb)


WE have a serious Exchange server disaster scenario. Exchng5.5 running on Win NT4 sp6 backed up by veritas.

we  can at this stage have only dir.edb.
can we get an empty system runninng with empty mail boxes  so that people can start sending and recieving mail ?

We also have about 100 field users.

We can recover priv.edb and pub.edb but will take time as the Priv.edb is about 60Gb and takes forever to restore. The Organisation can't wait.

How can we get an empty system running just with Dir.edb?
1 Solution
Make sure the Information Store service is stopped.
Rename the MDBDATA folder on all drives to MDBDATA.OLD.
Create a new MDBDATA folder wherever one was renamed.
Start the Information Store service and a new PRIV.EDB and PUB.EDB will be created.
FIRST:  FLIP YOUR MX's so SOMEONE ELSE gathers the inbound email.  (takes 30 seconds if you have good folks in place..)

SECOND:  DO NOT DO THE "DOUBLE-MAILBOX" thing.  Dear god, the nightmares never end.. especially with E55.  (more detail on this below)

Solution:  Tell your users and managers and VPs and Presidents that the restore *TAKES* 24 hours (or whatever it does) -- and a bigger budget for faster/larger/better backup hardware and processes ..  that seemed all too expensive before now kinda seems like it woulda been a good investment, no?

If you’ve ever pitched any kind of storage/backup solution to them before – now is the time to drag that out of the mothballs – and update it and get it implemented..

Solution:  Shut it all down, restore it and continue on.  

SOLUTION:  (let me write it again?)  -- DO THE RESTORE!

(take no middle ground on demanding appropriate hardware.  If they expect 24x7x365 uptime but want to run the 60GB db on a clone-P3 with a non-hot-plug array, or no array, quit now.  I regularly turn down $40/hr ($80k/year, approx) Canadian -- six to twelve month contracts for E55/NT4-2k3/ad-Migration stuff...  because they will not spend on the appropriate gear.  (nevermind that it’s worth more than that to get it done “right”..)  Stop risking your reputation because your VPs and managers of information-stupid are stupid.  Quit now.

Back to details;

I've done similar with a machine that had only about 200 users/15GB under E55 on it.  (we had limits on mailboxes, I’m guessing you did not..)  (have you ever used eseutil/d? -- did you know you're going to need ANOTHER 60GB free to get it done?)

If you just WIPE the Priv.edb and PUB.edb, or not -- you're going to get really roasted, one way or another.

I'd *REALLY* advise just shutting the damn thing off the the time it takes to restore the 60GB of data...   *IF YOU KNOW YOU CAN RESTORE IT*  

Let me explain how I learned about "CONFIDENCE IN RESTORING"..:

When I had to live through it – The above E55 machine tossed it (hardware bail – nice vendor Cpq/Proliant gear with a mirrored pair and 3x36-GB in raid-5..) – MB went, I think, and it went nuts.  We had a (tape) backup of our ~15GB priv and ~7GB pub on a tape loader.   That tape loader had all the stuff on it.  Way more than 100GB of files dumped on it too.  

So, backups took 22 hrs.  Done daily.  That’s big mistake #1.  You want to (operationally) keep your backups – to an OFFLINE media (required criteria) to less than 8-10 hours, MAX.  2-3 hours is what I target.  Writing to tape is slower than reading, generally..  

I think my HP/CPQ/E55 box pitched on about Dec. 21st or 22nd of whatever year that was.  

So, we diagnosed the hardware failure, sighed a breath of relief..  --   THAT I'D FOUND AND FIXED THE PROBLEM with the backup routine the week or so before..  and then stuck the tapes (6) (six-dlt-IV's that were 20/40GB's each..) and were happy that we'd be able to restore it over the holiday.  

During the season and panic -- I took a day (afternoon) to drive to my parents house (2 hrs away) and forgot to CANCEL THE PENDING, SCHEDULED BACKUP JOB.  Sure as doo-doo, it over-wrote the 3 day old set of data - leaving us with our next-closest backup to recover from -- which was about 3 months old.

So, FIRST THINGS when you deal with tapes:

1)  KNOW YOU CAN RESTORE (fire drill).  Just get other hardware and restore into it.  Use a PC, if nothing else.  If you don't, you're running a big risk.  Most of the risk is because you do not KNOW how to DO IT in a reasonable amount of time..  assuming the tapes read correctly, they actually have data on them, etc..  (no joke-- those are HUGE assumptions if you don't test 'em)

2) WHENEVER -- AND I MEAN EVERY TIME -- you pick up a backup tape, drive or set of tapes/drives to insert them into a system to RECOVER (read) from them -- FLICK THOSE GODDAMN LITTLE SWITCHES THAT MARK THEM READ ONLY.  YES, DO IT BEFORE YOU INSERT THE MEDIA.   YES, I KNOW WHY THEY ARE THERE.  

3) A 60GB PRIV.edb?  Unless you're running some sick 16-CPU/P4@3Ghz each machine, which is impossible for E55, to the best of my knowledge, ..   throw some limits on that store, brother!  Either that or migrate up versions – or blow it out horizontally, like web boxen.  I.E. 10 machines at $4000 each at 6GB each.


A)  Users with .OST files will be able to "sync" their data back up -- as long as they answer the magic question correctly.  (Remember, the NEW, BLANK PRIV.EDB is NEWER than their .OST -- so if they default sync, BYE-BYE too all that data..)

B)  Palm-Pilot users (see problem above..)

C)  If -- you get the 60GB store back up in an "offline" system, then you have to manually "read" all the data back into the production system while other data is in there.  (Exmerge?  -- Tape and band-aids and fingers crossed..?!)  For me, this generated nightmare problems.  My end-users didn't know if their calendar appointments were valid or not -- and literally didn't know where to go, who to call or what to do for several days.  

C)[pt2] -- if your users don't know where to go, what to do and don't know when the system will clean up their schedule for them -- when and if it gets sorted out, they'll come back and *TAKE A ***** on your head when it's fixed/done.  

No kidding, of the 200 users at the location my story came from -- I really thought, twice, that I was going to be physically assaulted.

(yes, I quit shortly thereafter..)

I *STRONGLY* recommend telling them the system is unavailable until it is fixed.

I know EE hates this, but call me at any hour, to discuss E55 recovery.  I broke that thing every way possible -- and a few more -- in the 3+ years that I ran it on 10 machines for 750+ users.

Scott, (416) 432-4334 (canada)

-- Best wishes, good luck.
nalinuAuthor Commented:
Hi Scott,

thanks for all the tech stuff. We were able to resurrrect the dir.edb and another which had all the common folders in it. We couldn't get the MS utilities to fix the Priv.edb. MS is crap !!!!!!!.

We used a utility from OnTrack www.ontrack.com which looked at the corrupted database and in seconds like magic showed us all the Mail boxes in it (including the mail of course).

we were then given the opportunity to connect the two priv.edb's (new blank and old corrupt) and suck the mailboxes one at a time to the other.

We also managed to store the mailfiles from all laptop users and sync them in. This saved address books calendars etc.

As you said the business now wants 100% uptime !!!!!!.

To cut a long story short we are now up and running again. A new server, tape drive etc are on the way.

EX55 is a Crap piece if software along eith it's utilities. Not good for enterprise computing.  

thank you so much again. How do I give you points.  I am new to Experts exchange.

Best regards and Many thanks.

Nalin Uduwawala.

PS: Aren't most users sick anyway ? Some have 2gb mail files and they are powerful.
PAQed with no points refunded (of 500)

Community Support Moderator
Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.

Join & Write a Comment

Featured Post

Get expert help—faster!

Need expert help—fast? Use the Help Bell for personalized assistance getting answers to your important questions.

Tackle projects and never again get stuck behind a technical roadblock.
Join Now