Re: Exchange 2003 Private Store Corruption

It could be that it is just DIMM 4 that is the problem. Do you have to have
DIMM's installed in pairs in this server? If not, I'd just pull out DIMM 4.

Also, can you post the entire contents of the error event? Does it mention
something to the effect of

Information Store (<PID>) The database page read from the file "<path to edb
file>" at offset <offset> for <value> bytes failed verification due to a
page checksum mismatch. The expected checksum was <checksum> and the actual
checksum was <checksum>. The read operation will fail with error <error
code> (<error code>). If this condition persists then please restore the
database from a previous backup.

and is the error code referenced a -1018 error? If so, then that *is* a
database corruption issue, not just a corrupt mailbox. Even if that is the
case, I still wouldn't recommend using eseutil or isinteg - those are really
last resort utilities. Moving mailboxes is one option. Restoring from
backup should be another option you should look at, or restoring from backup
to the Recovery Storage Group.

You might also look at the storage group properties and see if you have the
option "Zero out deleted database pages" option checked, and if so, you
might consider unchecking it.

Also, do you have any Exchange 2003 SP's installed? If not, once you get
this fixed, you should really make sure you are up to date. SP1 includes an
additional error correcting code that is designed to fix many of the -1018
errors that are as a result of a flipped bit.

Here are some KB articles that talk about 474 errors and -1018 errors.
http://support.microsoft.com/default.aspx?scid=kb;en-us;810411
http://support.microsoft.com/kb/327334/
http://support.microsoft.com/default.aspx?scid=kb;en-us;867626
http://support.microsoft.com/default.aspx?scid=kb;en-us;314917
http://support.microsoft.com/default.aspx?scid=kb;en-us;812531 - webcast on
understanding and resolving -1018 errors.

In your case, it really would appear that faulty memory is the underlying
cause.

--
Ben Winzenz
Exchange MVP
MessageOne
Read my blog!
http://winzenz.blogspot.com
http://feeds.feedburner.com/winzenz (RSS Feed)


"Corbin O'Reilly" <corbinoreilly@xxxxxxxxxxxxx> wrote in message
news:uj9tg.89246$qd2.10626@xxxxxxxxxxxxxxxxxxxxxxxxx
Thanks for the info Ben. Yes they are mounted now so hopefully you are
right and they will mount properly after the reboot. The HP Insight Online
Diagnostic Utility says "Correctable Memory Threshold Exceeded" for DIMM 4
so it does look like the memory is faulty. Have you seen 474 errors
before? In your experience what leads to a 474 error? The description
points to hardware. I am exporting mailboxes to PST files right now. Did
you see Mitch's previous response. I am just trying to get the server back
up and running without the faulty memory installed. My plan is to create a
new store, move as many uncorrupted mailboxes as I can to it, and then try
to repair the old stores.

"Ben Winzenz [Exchange MVP]" <ben_winzenz@nospamdotmessageonedotcom> wrote
in message news:%23iIUsWcpGHA.3600@xxxxxxxxxxxxxxxxxxxxxxx
Inline.

--
Ben Winzenz
Exchange MVP
MessageOne
Read my blog!
http://winzenz.blogspot.com
http://feeds.feedburner.com/winzenz (RSS Feed)


"Corbin O'Reilly" <corbinoreilly@xxxxxxxxxxxxx> wrote in message
news:OX7tg.8250$IF2.4659@xxxxxxxxxxxxxxxxxxxxxxxxx
Hello everyone. I need some help. We have an Exchange 2003 Enterprise
Edition Server running Windows 2003 Enterprise Edition R2 server. The
machine is a HP DL380 G4 server with 2 GB of memory and has been online
for about two months. The only thing running on this server is Exchange
2003 and Trend-Micro AntiVirus for Exchange. The server came standard
with 2 - 512 MB HP-brand DIMMs. We purchased additional memory from CDW.
They are 2 -512 MB EDGE-brand DIMMs and are supposed to be compatible.
On the first day that we powered up the server the server said there was
an issue with DIMM 4, the EDGE-brand chip. After a reboot the error went
away. About three weeks ago I noticed that the server had not completed
its online backup properly and two of our employees were having Calendar
issues. I checked the Event Viewer and it had tons of 474 errors which
says that there is probably a hardware problem. I created a new store on
the server and moved all mailboxes to the new store. The two mailboxes
with the corrupted Calendars would not move so I left them in the
corrupted store until I got back from vacation. While on vacation for
two weeks I checked the NTBACKUP logs daily and the online backups were
completing properly. Everything was fine for almost the entire time I
was gone. The day before I got back from vacation I got calls from 5
employees saying they were having strange Exchange problems. Some were
not getting new e-mails (we use cached mode), others had Inboxes that
were empty, etc. I checked the NTBACKUP log and sure enough the new
store I created is now corrupted too with a 474 error. My theory is that
the EDGE-brand memory chip is acting up and corrupting the store. What I
would like to do is shutdown the server and remove the 2 - 512 MB
EDGE-brand memory DIMMs from server, and boot it back up. I have a
couple of questions:

1. The server is partitioned into a RAID 1 OS C-Drive and a RAID 6
Exchange Data E-Drive. The server currently has 2 GB of memory and the
page file is located on the C-Drive and is set to Initial
: 3072 and Maximum: 6144. Since I will be removing the 2 - 512 MB
EDGE-brand DIMMs the new total memory size will be 1024 MB (1GB). Should
I adjust the page file to Initial: 1536 and Maximum 3072 in Windows 2003
BEFORE or AFTER I remove the memory?

I don't think it much matters whether you do it before or after, though,
except that if you do it afterwards, you will have to reboot a second
time.


2. Will the server perform OK with 1 GB of memory for a while? We have
about 75 mailboxes and a few public folders.

More than likely, it will be just fine. It doesn't sound like your
server is heavily loaded


3. Since both Private Info stores are corrupted will they mount properly
when the server is rebooted?

Are they mounted right now? If so, then once you reboot, they should be
able to mount. Dealing with corrupt mailboxes is somewhat different than
dealing with an actual corrupt database. Corrupt databases are not
mountable. Databases with corrupt mailboxes usually are mountable, but
the mailbox itself may or not be able to be fixed. I'd consider using
Exmerge to export whatever you can from the corrupt mailboxes. If they
can be moved, great. If not, then the mailboxes should be deleted and
re-create on the new store.


I am trying desperately to avoid being dead in the water tonight. My
concern is that the stores will not come up and we will be totally down.
I am getting more complaints from different employees so the problem
seems to be spreading and getting worse. I would appreciate your
expertise on this problem. Thanks.

As an additional note, have you run any server diagnostics to test the
new memory? What is your plan if removing the memory doesn't curtail the
problem? It's really better to have a true diagnosis that the memory is
faulty before you pull it out. HP should have some diagnostics that you
can run to test the memory. I'd do that first before you pull it out.








.

Relevant Pages

  • Re: Exchange 2003 Private Store Corruption
    ... I think you're on the right track with creating a new mailbox store. ... Moving mailboxes is one option. ... Online Diagnostic Utility says "Correctable Memory Threshold Exceeded" ... get the server back up and running without the faulty memory installed. ...
    (microsoft.public.exchange.admin)
  • Re: Exchange 2003 Private Store Corruption
    ... Once you've pulled the faulty RAM, create the new mailbox store and move ... then you can start working on the problem mailboxes. ... have DIMM's installed in pairs in this server? ... Online Diagnostic Utility says "Correctable Memory Threshold Exceeded" ...
    (microsoft.public.exchange.admin)
  • Re: Exchange 2003 Private Store Corruption
    ... Diagnostic Utility says "Correctable Memory Threshold Exceeded" for DIMM 4 ... Edition Server running Windows 2003 Enterprise Edition R2 server. ... I created a new store on ... Dealing with corrupt mailboxes is somewhat different than ...
    (microsoft.public.exchange.admin)
  • RE: server problems
    ... This newsgroup only focuses on SBS technical issues. ... >Thread-Topic: server problems ... >> web proxy service or SQL Server will normally use large memory. ...
    (microsoft.public.windows.server.sbs)
  • Re: Exchange 2003 Private Store Corruption
    ... Eseutil won't fix corrupt mailboxes, ... Edition Server running Windows 2003 Enterprise Edition R2 server. ... I created a new store on ...
    (microsoft.public.exchange.admin)