Well, I don't think that was the issue because those entries were deleted awhile ago and it was only one user with that PAB.
Since then, we've had issues with users having their mail stuck in the CAT queue.
This morning that was the case.
I disabled Antigen(which next to the Info store, was using the most mem). To disable Antigen, you have to stop all the Exchange services.
After the services started, mail was stuck in the CAT queue and the "Local Delivery" queue.
Exchange was in this state for about 5-6 hrs.
Now all the queues have emptied out and the "VM Largest Block Size" is up from 16mb to 30mb.
We get this issue ocasionally with the other mail servers(but not the Local Queue).
In my uneducated opinion I think a couple of things are causing this-
1.This corp does NOT set mailbox or mesage size limits. I've seen mail with 50-100mb attachments in the queue.
2.I suspect that maybe Antigen or one of its engines is at fault. Though maybe, its just using a lot of mem due to the size of the emails.
On the other hand, I've seen issues with just one or 2 emails of normal size sitting in the CAT queue for days.
This occurs on all servers. It is not isolated to one specific box. The only thing these boxes have in common is Antigen..
Thanks
On 6/6/06, Al Mulnick <[EMAIL PROTECTED]> wrote:
In that case, can you go ahead and show us the export of those pab entries that were found to cause the issue?Al
On 6/6/06, Tom Kern <[EMAIL PROTECTED]> wrote:Sorry,On the exchange server-"Smtp Server - Categorizer Queue length" is always at zero"MsExchange DSAccess Process- LDAP read time" is at zero"LDAP search time" is at zero as well.On the GC-"System- processor queue length" is at zero"PhysicalDisk(NTDS db/logs)- Avg disk/sec read" is at zero"Memory-available MBytes" is 533"Pages/sec" stays at zero but occasionally spikes to 90 for a sec.No errors on the NIC's on both Exchange or GC.Thanks
On 6/6/06, Tom Kern <[EMAIL PROTECTED]> wrote:Can you tell me what counters I should be looking at to determine GC perf?Thanks
On 6/6/06, Al Mulnick <[EMAIL PROTECTED]> wrote:One GC?Can you verify the performance on that GC? Waiting on a response from disk, GC, or other could absolutely cause the problems you are seeing.Al
On 6/6/06, Tom Kern <[EMAIL PROTECTED]> wrote:nope.I disabled Antigen AV and rebooted the box.Mail is stuck in the "local queue" and "messages awaiting directory lookup queue".In perfmon, the "VM largest Block Size" starts high(~80mb) but falls down to below 16mb in about an hour."VM Total 16mb Free Blocks" is at zero as is "VM Total Large Free Block Bytes".This is a Win2k standard sp4 server running Exchange Enterprise with sp3 and all rollups and hotfixes.We have 6 mailbox servers and one GC(maybe that could be an issue but I don't think it explains the "local delivery" queue issue).I spilt all the large groups up into smaller global groups.I followed this article, http://support.microsoft.com/kb/325044/en-us, but still no go.Any other help would be greatThanks
On 6/1/06, Al Mulnick <[EMAIL PROTECTED]> wrote:Another reason you'll get an Exchange consultant to recommend that is for management reasons. Few companies manage large groups well. Also, you can have better control over the expansion of groups with multiple separate groups, vs. one really large group.Tom, did you ever get good results?
On 5/31/06, joe <[EMAIL PROTECTED]> wrote:I am not aware of any limits in the size of DLs specific to Exchange. There is a recommendation to keep your DLs less than 1000 members. However, I expect that this is due to attribute ranging which in Windows 2000 was 1000 attributes and in Windows Server 2003 AD that is now 1500 members. The idea being that you can get all of the values in a single query instead of sending back asking for more over and over again. I did notice that Exchange does something odd when it has to start ranging to retrieve more members. It doesn't appear to be using the normal WLDAP32 library to do it. I was using Insight for AD from winternals and the additional calls to get the additional members weren't being caught, yet I could see them over the wire with ethereal meaning that the hooks that Insight puts into the WLDAP32 libs weren't seeing the calls... hence they weren't using the standard library.Breaking the users up into separate smaller groups and then nesting the groups is exactly what any Exchange consultant that came in would say.joe
From: [EMAIL PROTECTED] [mailto: [EMAIL PROTECTED]] On Behalf Of Tom Kern
Sent: Friday, May 12, 2006 11:15 AM
To: [email protected]
Subject: Re: [ActiveDir] Exchange queue(OT)
Well one thing I noticed is that the senders(and some recipients) are members of a AD security DG that has over 3300 members.I think the categorizer has a 1500 value limit for member?I'm gonna seperate the members into multiple local groups and then nest them into the DG.Maybe that will help.I'll let you know what I find.Thanks
On 5/10/06, Al Mulnick <[EMAIL PROTECTED]> wrote:
