Exchange Server Experts....

[Mslater]

Go an exchange server performing really badly.

The analysis returns High RPC, Disk BottleNeck and Memory BottleNeck.


High RPC is expected and is actually not that high for the hardware.

Just run tests on Memory and Disk -

SAS array certainly not hanging around, getting transfer rates approx 200Mbytes/Sec

Got 3Gb Memory free, no paging at the moment.

SO Exchange is lying !!!! Anyone got any bright ideas before I boot 53 users off the system.?

[Paulice]

Ask Mslater he's a computer wizz :D

[garlicfish]

Not touched Exchange for a long time but:
Which disk is showing the bottleneck? Datastore or Transaction logs?
How do you have your array configured? How many disks, type, layout, RAID level etc?
What sort of figures are you looking at that tells you there is a bottleneck?

[Gingerr6]

All system updates complete including SP for exchange? what version? how big is the database?

[Mslater]

Cheers Paul :)

There's no bottleneck its lying. I can get 500Mb data from the exchange database in just under a second using an extract query. just for reference its a SAS raid10 array of 6TB. The exchange data is 74GB.

System is fully up to date Exchange 2007 on Server2008 64bit of course with 32GB memory and 16 processor cores.

I've come to the conclusion exchange is sh1t. Funny but I came to the same conclusion the last time it played me up.

Oh well - better re-boot it.

[Rich_H]

Hi Mark,

Sorry didn't see the post was busy with issues of my own (pcounter)

Exchange 2007 is a nightmare, particularly when it comes to memory usage, It's a well documented MS <ahem> feature. Is it a single exchange box, ie contains the transport roles as opposed to mailstore/appliance seperate? I ask because I had a single exchange instance in a previous role where it was 'all in 1' and it did the same almost weekly, rebooting fixed it every time - in fact eventually i rebooted it on a schedule to avoid the issues.

I have 5 boxes where i am now, all virtualised though, 3 are back end - 2 front, doesn't seem to have the same issues.

Rich

[Mslater]

Yep it's all in 1. What really bugs me is even though you can manually restart each service it's no better until you reboot. So what it is that hides away is beyond me.


Rebooted earlier and it's all ok again, compared the stats and they're just about the same now as befor the boot.

So once against I have to conclude exchange is sh1t.

Can't even handle room resources properly since the last 'update'


However I can't complain to much. I'd be out of a job if it just worked. Lol

[danch]

Mark,
Can't you schedule an overnight reboot of the server as a stopgap?

[richieclarkie]

The last time I even touched exchange was 5.5 in around 2003.

Overall it was pretty stable but even then it would have a flid very occasionally! I do miss those days though :)

[Mslater]

[QUOTE=danch;844256]Mark,
Can't you schedule an overnight reboot of the server as a stopgap?[/QUOTE]

Not really, it runs backups over night, updates WSUS and antivirus plus its supposed to run the Exchange reorganise. It only just gets that lot done before the users hit it next morning.

Sunday night may be an option. Actually I think this might be quite good. I'll put a timer on the mains and test the UPS shut and restart every week.

Great Idea, hey Exchange is brilliant. (ignore that!)

[Tom Burgess]

Funny this, I'm currently troubleshooting a similar issue but on 2010. What size is your .edb?

Are your clients getting the "Retrieving data from server" bubble?

Are you using Microsoft Dynamics by any chance? RPC drop out is a daily grind in that instance.

I had a similar issue a couple of months back and I found that creating a new database and migrating the mailboxes to it cured it. I am loathed to do it again but if I don't have a resolution by the end of the day I will go that route.

An exchange server that is "up to date" is never a good thing in my experience :D

[Gingerr6]

[QUOTE=Mslater;844320]Not really, it runs backups over night, updates WSUS and [U][B]antivirus[/B][/U] [/QUOTE]

This could be the problem? what AV is it?

[heli]

As an exchange expert I could help here. I've seen what you described before and it's common on all versions even from 5.5. What your seeing is high usage on your database. Common issues are usually third party stuff that's bolted in to the MTA. Most of the time is antivirus and ripping it off helps other stuff is spam engines, fax or SMS stuff. It's almost impossible to work out what is hitting the DB so it's a case of elimination. Only once I've seen something hacking outlook cause the same issue, the virus was harvesting email addresses.