2.5 Million objects in A1 queue, and event 2850 in the eventlogs.

[ullabella]

Oki this is the scenario.

We have an Evserver that is archiving from one exchangeserver (Exch A)
A week ago someone moved half of all mailboxes from Exch A to Exch B
Exch B is located in a diffrent domain.

I was not avare of this, and i started to get alot of errors in the eventlogs, and services and taks crashing.

So i added the new domain to Enterprisevault.
And added Exch B as a target.

I sat the registry value SyncInMigrationMode as explained in technote
http://seer.entsupport.symantec.com/docs/284956.htm

So far so good.
The archives updated with mailboxes on Exch B

But we still have a problem.

And that is that the A1 MSMQ queue still holds about 2.5 million objects. And as soon as we start the archiving task against Exch A we get the error message 280

Whilst post processing a message an error occurred in CArchivingAgent::PostProcessArchivedItem() whilst calling CExchangeShortcutAccessor::BuildFromSessionAndLocation().
Error returned from BuildFromSessionAndLocation(): 80004005 Error returned from BuildFromSessionAndLocation(): %1

So how should i solve this ?

What would happen if i manuelly purge the queue ?
Will i lose some data ?

Thanks !

/

[Wayne_Humphrey]

You can purge A1 that's OK. A1 is Update Shortcut and Operation Failed.

What I suggest you do to start off with is set the arching times of both EX servers at different times.  So that EXA can archive, MSMQ is clean. then EXB archives.

Also, I would set the MSMQ Housekeeping Down from default 6 hours to 1 hour.  I would also consider reducing the number of threads on the moved EXB server down from 5 to 3, dependent on the bellow questions.

What is the bandwidth between the two sites?  What is the latency like too?

Set Message Cleanup Interval

The following fix installs a periodic cleanup of empty Message Queuing message files. The default interval for cleanup is six hours. This interval can be modified by a new registry key. To change the default interval, you must add MessageCleanupInterval to the registry.

1.    Start Registry Editor (REGEDT32.EXE).
2.    Locate and select the following registry subkey:
3.            HKEY_LOCAL_MACHINE\Software\Microsoft\MSMQ\Parameters 
4.    From the Edit menu, click ADD VALUE.
5.    In the Value Name Field, type MessageCleanupInterval.
6.    In the Data Type Field, select REG_DWORD.
7.    For the data value, enter the desired interval in milliseconds (for example, 3600000 (decimal) for one hour).
8.    Exit Registry Editor to save the changes.

The next time you restart the Message Queuing service, the new interval will be used to cleanup empty message files.


--wayne

[MarkBarefoot]

It sounds like the move was done in the middle of the archiving and post-processing operation. EV would be expecting to have the mailbox on the same Exchange server and you should be able to look at the items in the A1 queue to verify this.

Now that you have successfully sync'd these users and they are linked to the new Exchange server, these messages are effectively orphaned as they will be targetting the Archiving task for Exch Server A. You will need to run a cancel pending operation on the affected users, and get the messages re-archived (you will get a duplicate but this is unavoidable). You will then need to purge the A1 MSMQ and then associated Admin queue.

I would suggest opening a case with Support to assist you with this - that way you will get the right Technotes to help you.

Thanks

Mark

[ullabella]

GPKGuy, You are correct.
The move was probbably done in the middle of the archiving.

How do i run an cancel pending operation for all theese items without having to open each persons mailbox.
Can i just set the cancel pening items timeout in the mailbox policy to 0 ?
Then run the arhiving task in report mode against the Exch A ?
And after that purge the queue ?

And when you say the admin queue, which one do you mean ?

Thanks guys, you really helping me out here.
Very much appreciated.

[MarkBarefoot]

TN to clarify the cancel (you pretty much have it) - http://support.veritas.com/docs/273157 - solution 3

How to purge MSMQ - http://support.veritas.com/docs/276082

You should see an associated "admin queue" for each archiving task. When you purge any queue related to a task, these will go into the Admin queue for that task. If you don't purge this you will get lots of NACK messages.


Hope this helps :)

[jajensen]

The steps for canceling pending items are in this article:

http://seer.entsupport.symantec.com/docs/273157.htm

You want Solution 3 in this case.  Setting that Pending Shortcut Timeout value to '0' on your policy and running the archive task in report mode should get all of your items out of pending.

Note that the items have already been archived once, so you're going to get doubles of those items when they get re-archived.  The stores do single-instancing, of course, but those items will show up twice in the indexes.

[ullabella]

Oki and thanks all.

But when setting the Pending Shortcut Timeout to 0 and then run the archiving task in report mode against Exch1 it still spams the EV logs with event 2850.

Whilst post processing a message an error occurred in CArchivingAgent::PostProcessArchivedItem() whilst calling CExchangeShortcutAccessor::BuildFromSessionAndLocation().
Error returned from BuildFromSessionAndLocation(): 80004005 Error returned from BuildFromSessionAndLocation(): %1

And after some time ev will automaticly shut down the ev services due to the large amount of entries in the eventlog.

Suggestions ?

Should i skip this step and just purge the queue ?

[TonySterling · Accepted solution]

you should purge the queue and then run the archive task in report mode to reset the pending items.

[JustMy2Cents]

Hi,

I have similar issue in the Vault I'm administering. I looked at the documents posted here and they seem to be for EV5.0 up to 2007 SP2.

I'm using EV 2008 SP2 so does these apply to that version also?

http://seer.entsupport.symantec.com/docs/273157.htm

http://support.veritas.com/docs/276082

 - 0,02$

[MarkBarefoot]

They do apply to EV8 as well, as the MSMQ part is Microsoft anyways. It looks like the Technote wasn't update to reflect this.

[ullabella]

Thanks guys for all help.