Log files growing after failed backups

Have a full backup of your store before you defragment, just in case things go wrong.
Leave the logs as to where they are.
Once the defrag is complete, you need to immediately run a full backup, as your database files will have been changed.

Rajith.

Since the database files will be changed after the defrag, can/should I at that point move the log files off to another storage device?

Also, since the database mounts, do you think I need to do an eseutil /p before running the defrag? I read a few articles that mention doing so and then running isinteg too, but it seemed they were referring to databases that were so corrupt they would not mount.

HI!

As mentioned by Rajith it doesnt make sense to take full bakcup, since the NTBackup itself is failing.

please make sure the stores are mounted. If the stores are mounted and still you get the transaction log files growing than stop the SMTP service check log files.

If still you are getting log files growing then database is corrupt.

Also check if you have Journalling is enabled on your server. Also try to disable journalling.

Let me know.

Thanks.

Jawad.

The stores are mounted. I tried stopping a few services. First, I stopped Groupshield. No change. Next I stopped both smtp and pop3 virtual servers (we use an off-site message archiving solution that pulls all our emails via pop3 (no one else uses pop) - to do so we do have journaling on). I still noticed log files getting created faster than would be expected with little to no email activity (most of our users are not online today). So, I started both smtp and pop3, and was going to turn off journaling when I noticed the creation of log files had slowed to a pace one would expect under the circumstances. I watched it for a 30 minutes or more and the log file creation stayed the same, much slower that is. Not sure why stopping and then starting those services would do that (as I said, with them stopped (before starting them back up) the log files grew at the same rather rapid pace). I just started the Groupshield service too, and the log file growth remained slow. Does not make a lot of sense since rebooting the server the past two times (granted it was a typical work day when that happened) did not alleviate the problem.
I am still going to proceed with the offline defrag of the mailbox store, and time permitting, the public store. My remaining question is whether I should do a repair first (eseutil /p) or just go right to doing the defrag. Any thoughts on that? (and any thoughts on anything else I described in this reply)

Hi!

i would not recommend to go for "eseutil /p" , you can go ahead with offline defrag and isinteg.

make sure that before running defrag, check the dsik has 110% free space. and once defrag is complete then mount and dismount the store and run ISINTEG.

let me know, if you need any help.

Thanks,
Jawad.

Thanks Jawad. I will start with the defrag and proceed as you suggest. Just curious, any particular reason why you do not recommend eseutil/p? By the way, after completing these tasks would a full backup remove the log files or will I need to move them off manually?

Thanks Jawad. If I start with eseutil /g and find corruption in the database should I run isinteg before doing the defragmentation?

HI!

If you find that the database is corrupt after running "eseutil /g" than will have either restore from backup or if backup is not available than repair using eseutil commands.

I am sure the database will not be corrupt.

Note: eseutil /g might take some time depending upon the size of the database.


Let me know.

Thanks.

Jawad.

The priv.edb is 25GB and the .stm is 6GB. If the eseutil /g runs as long as the defrag I may just start with it and skip the /g. I tend to think you are right about not being corrupt. I just do not see errors in the event log or other anomalies when mounted to make me suspect corruption of that magnitude. I ran a defrag on our public store a few months ago (it was 43GB or so) and it took about 5 hours or so.

yes, definitely we can skip eseutil /g becasue it will tae the same time.

let me know if you have any issues.

(make sure that before running defrag, check the dsik has 110% free space. and once defrag is complete then mount and dismount the store and run ISINTEG.)

Thanks.

Jawad.

Sounds like a good plan. I have 58GB of free space, so I am in the clear on the 110% requirement. I take it the syntax to be used for isinteg is: isinteg -fix   correct?

Hi,
Wait......................!! Why do you want to run ESEUTIL, obviously, if the backups have failed, then Log files would continue to grow. The log files growing does not indicate necessarily, that EDB is corrupt.
Please share the following information:

• What is the Edition, that you have of Exchange 2003 - STD or ENT
• What is the exact error that you are getting
• How fast are the log files growing.

Thanks
Nitin Gupta (gupnit)

Hi!

The Syntax fot ISINTEG: isinteg -s servername -fix -test alltests.

please let me know if you have any confusion,

Thanks.

Jawad.

Gupnit, I have Exchange 2003 Ent (SP2). After the backup fails and before I either rebooted or restarted the IS, the errors are as follows: "Database error 0xfffffd9a occurred in function JTAB_BASE::EcUpdate while accessing the database "First Storage Group\Mailbox Store (ACXCHGVS)".
and
"Database error 0xfffffd9a occurred in function JTAB_BASE::EcSetColumn while
accessing the database "First Storage Group\Mailbox Store (ACXCHGVS)".

These two errors repeated numerous times after the backup failed.
Anew log file was being created almost every 2 seconds or so (and that with no one on the system and little to no email activity). As noted above, once I stopped (I waited a few minutes before starting again) the smtp and pop3 virtual servers the log files continued to grow, but after restarting them it dropped off to one log file taking about 4+ minutes to be generated.

A few weeks ago I planned to do the offline defrag today because I had not done so in a while (cannot remember when last - 1yr perhaps?) and our users had cleaned up their mailboxes recently too.

Hi,
Also for your ready reference: http://support.microsoft.com/kb/905801
Thanks
Nitin Gupta (gupnit)

HI!

I am sure my friend might be knowing the diffrence between the offline defragmentation and moving mailboxes.

This will only remove the corruption but will not shrink the database.

I have clearly mentioned that we dont need to run eseutil commands, but the author wants to go for defragmenation. So i helped him with some basic steps.


Thanks.

Jawad.

@ Jawad - Nothing against you gentleman, unfortunately, I would not agree with you on your previous comment. It is a industry wide practised method if you have Enterprise edition and you can verify it with anyone you feel like :-) ! See Offline Defrag would help you with White SPaces primarily but would not reduce the actual size of EDB, but creation of new Database and Moving Mailboxes would do everything for you.
Cheers
Nitin Gupta (gupnit)

Okay, let's see if we can come to some agreement on the best way to proceed. One of my goals is defragmentation, but a clean, properly functioning database is my top priority. So, is creating a new database and moving mailboxes to it the best approach for that objective? If so, does doing that adversely affect OWA, or OMA for instance? Or, since we only have one server and the new mailbox database will be on it too, there will be no issues for clients that connect? Also, what becomes of the current database? Will it remain the same size and so deplete even more the free space on my server?  

Hi,

1. Create New EDB on same server, it is definitely, the best approach
2. It will not affect Client Access (OWA/OMA etc) in any way
3. Once Mailbox Moevment is over, backup is done fine, then go ahead and remove the old Mailbox database.

Generally, I do not enter open threads, but if I feel, something can be corrected or I can contribute a little extra, I intervene. Rest is your call :-) !
Thanks
Nitin Gupta (gupnit)

Hi!

As per my knowledge i practice the database everyday. and when it comes to database maintenance i would suggest you to with the defragmentation. since moving mailboxes will remove corruption if present any and in doing so you might loose some emails which might be important to your users.

So the best way would be go for offline defragmentation. Both the methods will take same or some less time.

the online maintenance maks the white spaces and offline defragmentation removes it.

The choice depends on you which you choose.

Thanks

Jawad.

My Last Comment on this topic -

• Offline Defrag means your Mailbox DB would be unavailable to user for the period of time, which means downtime. Mailbox Move would only affect 1 user at a time, and that too for a fraction of a second.
• One Offline Defrag vs Mailbox Move: http://www.petri.co.il/forums/showthread.php?t=25356

Cheers
Nitin Gupta (gupnit)

Thanks to you both. I understand there usually is a variety of approaches when it comes to such things and I appreciate your suggestions for both methods. I will soon select (soon since in about three hours I will begin this task of fixing the DB) the method I will use. Before doing so, I have a few more questions for gupnit on the move mailboxes method: 1) I assume the move generates more log files, and if so, I am concerned about space (I have about 58GB free right now). I could move the log files (about 20K of them) to a NAS drive to free up space; 2) should I create a new folder for the new store so the log files for it are in a different location?

I have started down the path suggested by gupnit. I read the post in the Petri forum listed in gupnit's last response and saw a reference to rebooting the server once the now empty (except for system accounts) mailbox store has been deleted. This seemed related to having created a new storage group, which is not what I have done. I created a new mailbox store within the first and only (default) storage group. So, do I need to do this after removing the empty mailbox store or not?

Great ! Once you have done the Mailbox Move and restarted the server, go ahead and delete the old DB, and I would say reboot again. You should be fine, this is a pretty straight forward process.
Cheers
Nitin Gupta (gupnit)

gupnit, there is one mailbox (for the domain administrator account) that will not move to the new db. It generates the following error (see below) every time I have tried (I am okay with deleting it and recreating it in the new db (it has a dozen or so emails which I could export/import).

Also, there are three mailboxes that while they moved to the new store still show an entry in the current DB (my mailbox is one of them).

Error during move:

 <?xml version="1.0" encoding="unicode" ?>
- <taskWizardRun taskName="Move Mailbox" dcName="ACCAPPS" buildNumber="7638" runningAs="Administrator@ACCONTROLS.LAN">
  <timespan startTime="2009-05-03 12:31:17.304" milliseconds="16328" />
- <moveMailbox mixedMode="false" maxBadItems="0">
- <destination>
  <database>/dc=lan/dc=accontrols/cn=Configuration/cn=Services/cn=Microsoft Exchange/cn=accontrols/cn=Administrative Groups/cn=accontrols/cn=Servers/cn=ACXCHGVS/cn=InformationStore/cn=First Storage Group/cn=Mailbox Priv2 (ACXCHGVS)</database>
  </destination>
  </moveMailbox>
  <taskSummary errorCount="1" completedCount="0" warningCount="0" errorCode="0x00000000" />
- <items>
- <item adsPath="Administrator" class="user">
  <progress code="0" milliseconds="16328">Opening destination mailbox.</progress>
- <summary isWarning="false" errorCode="0xc1050000">
  The information store could not be opened. The MAPI provider failed. MAPI 1.0 ID no: 8004011d-0289-00000000
- <details>
- <source>
  <database>/dc=lan/dc=accontrols/cn=Configuration/cn=Services/cn=Microsoft Exchange/cn=accontrols/cn=Administrative Groups/cn=accontrols/cn=Servers/cn=ACXCHGVS/cn=InformationStore/cn=First Storage Group/cn=Mailbox Store (ACXCHGVS)</database>
  </source>
  </details>
  </summary>
  </item>
  </items>
  </taskWizardRun>

Hi,
When you do a move , please increase bad items list from 1 to around 10 and then retry. Also refer: http://support.microsoft.com/kb/278966
Thanks
Nitin

HI!

First do you see the event id 9168 and 1008 in the application log ? could you please paste the information.

Also let me the Domain Admin account belong to which OU? if other than Users OU, try moving it to ther USers OU and then move.

Let me know.

Thanks.

Jawad.

HI!

First try moving the Domain Admin Account to different OU and move mailbox.

You could also try rebooting the server once. I believe after rebooting you would be able to move mailbox.

 Cheers,

Jawad.

I was running a backup and a few other things and only now got back to the error when moving the mailbox for the administrator. I tried increasing the count of bad items to 10, but still failed. I see in event viewer error 1008:
Unable to move mailbox 'Administrator'.
Error: The information store could not be opened.
The logon to the Microsoft Exchange Server computer failed.
MAPI 1.0
ID no: 80040111-0286-00000000
Also, error 9167:
Unable to get an administrative interface to the mailbox store on server 'ACXCHGVS'.
Error: The information store could not be opened.
The logon to the Microsoft Exchange Server computer failed.
MAPI 1.0
ID no: 80040111-0286-00000000
Also, 9168:
Failed to open mailbox '/o=AC Controls Company, Inc./ou=ACCONTROLS.COM/cn=Recipients/cn=Administrator' in mailbox store '/O=AC Controls Company, Inc./OU=ACCONTROLS.COM/cn=Configuration/cn=Servers/cn=ACXCHGVS/cn=Microsoft Private MDB' on server 'ACXCHGVS'.
Error: The information store could not be opened.
The MAPI provider failed.
MAPI 1.0
ID no: 8004011d-0289-00000000
And, error 9175:
The MAPI call 'OpenMsgStore' failed with the following error:
The information store could not be opened.
The logon to the Microsoft Exchange Server computer failed.
MAPI 1.0
ID no: 80040111-0286-00000000

I rebooted the server about 30 minutes ago and the move of this mailbox still fails. Yes, the administrator is in the same OU as all the other users. Any problem with simply deleting the mailbox and creating a new one for the administrator user?

Hi,
What is this administrator user you are talking about?
Thanks
Nitin

Hi!

If you have taken PST of the Domain Admin , than you could go ahead and delete that mailbox. but before you do that confirm whether the PST taken is good and has all the emails present. simply connect it to your mailbox and verify whether it has all the emails.

You could delete mailbox and create a new mailbox for that user,

Cheers.

Jawad.

This is the domain administrator account. Yes, I planned on exporting the mailbox contents to .pst (very little to export) and then deleting and rereating the mailbox. Since this mailbox has not moved I still have not deleted the initial mailbox store. I was able to get my first 'full' backup of the store yesterday afternoon, so that is good.

HI!

Great, definitely we seem to be on right track.

If you need any help, please let us know.

Thanks,
Jawad.

Sure, go ahead :-)
Thanks
Nitin

Well, my second full backup took a little over 12 hours to complete (usually takes half that time or less than half). The resultant .bkf file is 89GB whereas the pub1.edb is 44GB and the priv.edb is 19GB. I do not know why that is the case, it does not add up. Log files seem to be spawned quicker than I would expect (1 or two a minute). I did have journaling set up and pointed to a universal distribution group that consisted of two members that had separate mailboxes (one for a contact management system we use and the other for a hosted message archiving solution). Prior to this we have had journaling on for 2-3 years with no problems, but then it was only pointed to a single accounts mailbox not a group. It seems since going with the group the backups did take a jump in size (about 9GB at first and then up to 25GB at most before these problems surfaced). This morning I changed the journaling feature back to using only the single account as before. I mention all this in the event it helps answer the question as to why the backup is larger than the sum of all the database files. Regardless of the answer to this question I will assign points for the answers to my prior questions. Thanks again to you both for your assistance.

It would be a good idea to have JOurnal Mailbox kept on a seperate Database, as it would have a copy of each any every mail
Glad to help
Cheers
Nitin Gupta (gupnit)

Since I have yet to remove my first mailbox store (because of the admin account I could not move), would it accomplish what you are suggesting to move the journal mailbox back over to that store (it would be the only mailbox in it)? After moving all the mailboxes this weekend I dismounted the initial mailbox store and used eseutil /d to reduce its size for the backup, so that mailbox store is fairly clean.
If it makes any difference to your suggestion, our contact management system moves all the emails from this journal mailbox into subfolders in the public store.

I am not sure of your contact management system, so I would not be in a position to make a recommendation in that regards, I would suggest you talk to the CMS vendor for that part :-)
Cheers
Nitin

I am posting this as a closing comment and a caution to others who may experience similar problems. This morning I remembered that I had turned on envelope journaling (using the MS supplied exejcfg util) for use with the archiving service we were trying out. So, I turned it off and guess what . . . my log file generation dropped from 1 every few seconds to 10 per hour! The performance on the Exchange box was noticably better. Users stopped getting 'retrieving data' messages and the hour glass when moving around in Outlook. All that remains is to get my backups on track again and all will be normal once again in my Exchange world.
Thanks to gupnit and Jawad for your assistance.