service pack upgrade. Empty msmq?

[total_noob]

Im currently planning on upgrading Enterprise Vault from the original release to sp4.

I have the upgrade documentation ready but im wondering about one part of the upgrade process which is emptying the MSMQ queues.

How do i do this?

We have about 750 users. The MSMQ queues are currently on 60 in the A5 queue and 2500 in the EV storage archive queue for exchange.  Also 20 in the restore spool.

Ive read something about setting the archive task in report mode overnight. Will this help?

[Nick_White]

You need to allow the storage archive queue to finish first as those are items which are pending in the mailbox but haven't been commited to disk in the Vault Store partition. When they are commited (your storage service might be in back-up mode which would also explain the restore pending items) the Archiving task will be used to convert the items from pending to shortcuts, so that will also need to be in normal mode. The A5 queue can have items in it as it's the queue of mailboxes that are due to be processed, so there wouldn't be items caught mid-archive for those entries. Once the queues are clear you can then stop the tasks and services and proceed with the upgrade

[GertjanA]

Hi Total Noob

Putting the task in report mode, then let it run overnight will allow the queues to empty cleanly.

If not putting the task in report, you run the risk that queues keep being filled. I always placed task in report mode the evening before the day of the upgrade. (on a larger environment!) 1 night should be sufficient.

[total_noob]

Thanks for quick answers.

So what ill do now is putting the archive task into report mode over the weekend and upgrade EV on monday.

I guess this will empty the queues then.

[GertjanA]

Hi TN.

Weekend should be more than sufficient.

[JesusWept3]

I'm gonna go out on a limb here and say if you have 2500 items in your Storage Archive queue, they are not going to go down because you've breached the 1GB MSMQ limit. Storage Archive will always process regardless of tasks schedule or site schedule UNLESS EV is in a read only/backup mode

You should see the numbers start to go down, but if you don't within an hour, what i would do is browse to your MSMQ location and check the size of the directory, if its more than 1GB you are going to want to go to the properties of your MSMQ in Computer Management and up it to a larger number, it will require a reboot but then your items will start moving

The reason  for this is though that where as most queues (such as A1, A5 etc) are just single lines pointing to a mailbox or a message, the items in Storage Archive, Storage Spool etc are exact copies of the message itself, so if you archive a 5MB email, you will have two messages in Storage Archive, 1x3MB part of the email and 1x2mb for the rest of the email

The chances of each message being small like 5k probably are quite small

[Mohawk_Marvin]

Just to throw a spanner in the works... Isn't it EV best practices to have an 8gb MSMQ location?

Otherwise though JW2 is right 2500 items in Storage Archive probably means EV has fallen over somewhere.

[JesusWept3]

I dont think theres anything in best practises regarding setting the storage allocation, as far as the DS is concerned it looks at MSMQ Permissions, locations and Memory settings (KernelMemThreshold and Page Pool Size) thats not to say it isn't out there that it should be set to 8GB, i'm just not aware of it :)

[AndrewB]

i've been told that by symantec support

[JesusWept3]

yeah but theres a difference between what supports best practices are and what official best practices are

[total_noob]

Sorry to bring this thread up again, but things have not evolved well.

The archiving task has been in reporting mode all weekend, but it did not help much. The schedule for the task is just 2 hours each night because of colliding backup operations.

The A5 queue went from 60 to 45.
The restore spool queue still got 10 in queue. Ill ignore these messages because its a known error mentioned here:: http://seer.entsupport.symantec.com/docs/316588.htm

The storage archive queue is still aprox. 2500. It actually went up with 10 messages.



The MSMQ size was already set to 8gb. 
Backup mode is off and i deleted the old legacy backupmode registry keys today as well.
The disk with the vault store partition has about 45gb free space so it should be enough space for the archiving task.

There are no error messages about this so im not even sure there is a problem here..

I suspected that maybe the site schedule and its timeframe was to low, but as jesus mentioned; archiving are not affected by scheduling.


I called symantec and they suggested a reinstallation of MSMQ. 

[Mohawk_Marvin]

Clear the solution flag, or maybe start a new thread to get attention to the issue.

If Symantec support have asked you to re-install MSMQ thats where you should probably start  :)

[JesusWept3 · Accepted solution]

reinstall msmq is a really really horrible idea, it might fix it but its shotgun at best
how big is the MSMQ area at the moment? i mean it is possible that it has gotten above 8GB if journaling is enabled on a queue somewhere

honestly you'd be best dtracing storage archive and then restarting the storage service to see what happens when it attempts to open the queue and process a message

[total_noob]

No journaling enablet.

The MSMQ is not above 8gb. Around 800-900mb actually.

I will run a Dtrace and see what happens. 

[Mohawk_Marvin]

Is MSMQ installed without AD integration?

[total_noob]

Yes, im pretty sure its not AD integrated