Re: Jet_errMissingLogFile

From: Dave Howe [MSFT] (daveh_at_online.microsoft.com)
Date: 06/21/04 

Date: Mon, 21 Jun 2004 07:47:45 -0400

On Fri, 18 Jun 2004 15:49:34 -0400, "Yor Suiris"
<yor@hallgroupNOT.net> wrote:

>But a question, before I started this and during there were comments about
>needing the logs as well to do a restore. I had made sure I had the Exchange
>logs. But now I think those were not the logs they were talking about.
>Also on the instructions about backup\restore there were comments about
>rolling in or out the transaction logs.

Hi Yor,

Whether log files are essential is determined by how you plan your
recovery in the even of a dirty shutdown. But before I go into that,
let's talk about log files.

Exchange is an ACID compliant messaging system. This means that
Exchange uses a database and log files for storing mail, and the log
files provide means of recoverability from a dirty shutdown. A dirty
shutdown is where something happened to the server (ie. power failure)
that caused the databases to dismount without first committing all
outstanding transactions into the database (inconsistent).

Let's assume that your server lost power and the databases shut down
inconsistent. When you power it back on, ESE Logging/Recovery will
occur as the Information Store service starts up. The header of the
database (ie. eseutil.exe /mh c:\program
files\exchsrvr\mdbdata\priv1.edb) tells ESE that there are outstanding
log files to be replayed (Logs Required value). Then ESE looks up the
last committed log entry (ie. eseutil.exe /mk c:\program
files\exchsrvr\mdbdata\e00.chk), and determines that this is where ESE
needs to start replaying log files (Checkpoint value). ESE
Logging/Recovery will take over and begin committing those
transactions up to E00.log (the most recent log entries), and when it
finishes, the database will mount successfully. This is what's
referred to as a "Soft Recovery" ... the replaying of log files to
mount a database.

Let's now assume that on Tuesday you did a full backup of your
Exchange databases, and on Wednesday the hard disk that holds your
database fails (not the log files, those should be on a seperate hard
disk). You've now replaced your hard disk, and now you need to
recover Exchange. How do you do it? Well, if you have all the log
files on a different disk, then it should be as simple as restoring
your last backup from tape to the new disk. After restore completes,
it will roll forward replaying transactions from the log files into
the restored database, and when finished the database will mount. As
you can see, ESE Logging/Recovery during a restore operation behaves
very similarly to a dirty shutdown situation.

OK, let's shift gears a little here and talk about a "Hard Repair" of
an Exchange database. Let's assume you've an administrator who has
never heard of the word "backup" ... ;) Exchange is running fine, and
you've never had a failure, so... Who needs a backup, right? Well,
on Friday you check your application log on your Exchange server and
find that you've been generating -1018 events for quite a while. If
you're not familiar with those, you should see the articles I've
referenced below. Since you don't have a backup, you're left with a
three stage repair operation - repair, isinteg, and defrag. Those
operate typically at 4 gigabytes an hour, so you can imagine how happy
your boss will be with you for taking the server offline for days
while you run a repair operation on a 70 gigabyte database. The bad
thing is that every time you do one of these repair operations, you
risk data loss. And, any log files that you created can no longer be
used for recovery since everything within the database will be
rearranged. Pointers within the log files to tables/rows within the
database will no longer be valid.

314917 Understanding and analyzing -1018, -1019, and -1022 Exchange
database
http://support.microsoft.com/?id=314917

259851 XADM: Ramifications of Running the ESEUTIL /P or EDBUTIL /D /R
Command
http://support.microsoft.com/?id=259851

I hope this helps to explain some of how Exchange operates on the back
end. I'm curious to know whether you were ever able to get your
mailbox store back online.

Have a good day! 

---
Dave Howe
Microsoft PSS
This posting is provided "AS IS" with no warranties, and confers no rights.

Relevant Pages

  • Re: Backend Exchange migrate to New Hardware (Server)
    ... Exchange mailbox stores and Exchange public folder stores. ... When you use Backup to restore Exchange databases, ... Storage Engine to restore Exchange database files and their ...
    (microsoft.public.exchange.setup)
  • Re: Exchange server
    ... microsoft exchange information store service or both. ... you unable to mount the exchange database with error c1041724. ... The Eseutil log files have not been removed. ... Programs, point to Microsoft Exchange, and then click System Manager. ...
    (microsoft.public.windows.server.sbs)
  • RE: Exchange restore on SBS2003 from SBS2000 backup
    ... I understand that you would like to restore ... the Exchange database from the SBS 2000 backup to the SBS 2003 Server. ...
    (microsoft.public.windows.server.sbs)
  • Re: Exchange server
    ... Determine if the eseutil /p command was run on the database. ... These commands assume that you are running Exchange 2003 from drive C ... All the log files have been committed to the information ...
    (microsoft.public.windows.server.sbs)
  • Re: update: HELP EXCHANGE IS DOWN - RESOLVED
    ... If you'd had an unrepaired issue that caused the database corruption, you might have restored the PST files, only to find yourself right back in the same place. ... I haven't had a major Exchange issue in a few years now, but when I have, my primary consideration has been whatever will take the least time to get us back in business, regardless of cost. ... From what it looks like the logging/recovery failed and I guess its time to restore from backup. ...
    (microsoft.public.windows.server.sbs)