Has anyone else noticed a memory leak in Store.exe on a new installation of Exchange 2007? After about 3-4 days Store.exe consumes a large amount memory. If I move the cluster group to another node, that node will also experience the same issue within a few days. Are there any patches for this, or have I not configured something correctly?

That is called "a feature". Seriously, store.exe will consume 70 - 95% of all physical memory in the server. So, if your server has 8GB of RAM and you have hundreds or thousands of users, store.exe may consume 6 or 7GB of that RAM. This allows the database engine to more effectively cache the recently used data. Now, if you are seeing more than the physical memory being allocated, then that might be a problem.

I understand that Exchange will use a large portion of the allocated memory, but there are currently only 130 users on this server and the database is fairly small. We are in the process of migrating the entire company off GroupWise and I'm a little hesitant to move a couple of thousand users onto this cluster. The server has 16G of physical memory and a 25G pagefile. As of now, store.exe is using 14.4 gigs of RAM. Doesn't this seem a little high?

I have also found that the store.exe will consume ALL available memory and Page file resources. I have 2 clients with the same issue. Server eventually crawls to the point of not being able to deliver mail. Rebooting clears the memory leak. on one server it had 10GB of page file in use and 6GB of RAM. I don't think this is a 'feature'. I was about to open a case with MS.

Update: It appears that our problem might be related to the size of the Page File on the OS. We have changed the Pagefile settings to use a 20GB page file. This appears to have helped the issue greatly.

There has been a lot of debate about the size of the page file necessary for Exchange, but in general the initial size of the page file should be about RAM+100MB.

We currently have 3 clients running Ex07. We increased the pagefile on all 3 to 20GB. Here are the results of usage after 24hrs since change- Server A - (6GB RAM, 20GB PF) - PF usage - 6.9GB, store.exe 3GB, 500mb free physical ram. Server B - (6GB RAM, 20GB PF) - N/A recently rebooted. Server C - (4GB RAM, 20GB PF) - PF usage - 3.5GB, store.exe 1.8GB, 200mb free physical ram. Server A and C appear to run much better. Server B is still having issues, but now with Edgetransport.exe and SMSUtilityService.exe using the bulk of memory. To the point that mail flow slows and stops. Resolved by stoppingand restarting both services. (SMS = Symantec Mail Security for MS Exchange v6.0.0.177) Server B receives about 50,000 emails a day. 95%of it is spam. -I've seen a PF size recommended as RAM+10MB on Technet. I assumed the 10MB was a typo, since that size makes such a little difference. +10GB makes more since, but still seems like a lot. But, even +100MB seems such a small amount to add.

You can change an attribute for the server in ADSIEdit.msc in the Windows Server 2003 support tools. Once in ADSIEdit, go under: Configuration -> Services -> Microsoft Exchange -> (Yourdomainhere) -> Administrative Groups -> Exchange Administrative Group -> Servers -> (Yourserverhere) -> Information StoreRight click on Information Store and find the attribute for msExchESEParmCacheSizeMax and edit it. It's not apples to apples so you'll have to play with it. 147456=1.2GB so good luck. It worked on our Exchange 2007 Server installation.

I am experiencing this issue also. Is the adsiedit change the final word ? Is there a patch to be brought out to fix this ?

I have the same problem (SMSUtilityService.exe) and have had it since day one (about 5 months ago). i have to restart the services almost daily. IS THERE A FIX TO THIS PROBLEM!?

I have just finished an Exchange 2007 installation and I am running into the same problem. Has anyone had any success in fixing this issue besides scheduled reboots?

Same issue here. We have only moved about 300 of the 1300+ users over to the 07 server and it already seems slow. The store is constantly using 30/32 GB of the RAM. Hopefully SP1 will offer some performance increases as well.

Same here. Any solutions yet. New server with 4gb and only 256 mb is available. Is this because of Symantec mail security?

We have had similar issues since moving to Exchange 2007. We have patched it with all currently available/relevant hotfixes/rollups and still no change. We only have110 users, 8GB RAM installed and a current pagefile set at 16GB. Within 12 hours of a reboot we get the following event in the event logs and have to reboot again, Physical memory load = 93% (limit is 94% to start dehydrating messages.) Just starting to model some performance statistics on the issue and its store.exe which appears to be leaking. Interested in any suggestions anyone may have in narrowing down the root cause...nothing of relevance in any Microsoft KB articles yet.

Yes I must agree now with Jim McBee, this is a "FEATURE" from Microsoft to keep us on our toes at all times. I have found that the memory usage will fluctuate to extreme lows but does recover. However because of this nice little feature I have encountered some problems just attempting to open the command shell because of lack of memory. Maybe this is another feature from Microsoft to only allow configuration changes until we have had plenty of rest waiting for the memory to recover to a useable value.

We have exactly the same situation: two separate clusters of 3 Store-Servers each and two Access/Hub Servers for each cluster. The Access/Hub Servers are Ok. Both clusters had 8GB physical RAM on each machine and as we experienced the increasing consumption of memory on one cluster with roughly 12'000 mailboxes, and having to restart the store.exe or even reboot the machines, we added another 8GB of memory on each server of this cluster last Monday. After 48h since I rebooted the machines, I am not sure, how to interprete this screenshot. But I fear really Exchange is behaving wrong or maybe the only third party products installed and running on the same servers, which is McAfee Enterprise and McAfee GroupShield for Exchange. On the other cluster we have only about 2'500 mailboxes and even if PF Usage (as displayed in the task manager) fluctuates inbetween 7 and 8.5 GB, the cluster behaves more stable than the other one. On all machines Virtual Memory (Page File) is configured to standard values of 2 to 4GB. I think that even increasing physical or virtual memory should not help a lot because already Exchange 5.5 and 2003 run stable with a lot less of memory. The reason has really to be on the OS/Exchange or perhaps on the Exchange-Virus-Scanner side. Has anybody investigated further? Rosario

We are having the same issue. The strange thing is that we only have migrated about 20 users toa box with 8 gigs of RAM and only after two days the system is usingclose to 90% of RAM. I applied SP1 and I am still having the same issue. Anyone have any luck with this or is there a serious memory problem. What is going to happen as the IS grows larger? Does it get worse?

No offense intended to anyone running an AV product on the mailbox servers, but doing that is a bad idea. Run the AV scans on your hub transport servers, all of the mail traffic passes through there anyways.

Exchange will consume available memory in an effort to create an efficient cache to improve performance. The memory utilization you're seeing is completely normal and is by design. If other processes require memory, Exchange will reduce the size of the cache, reducing the amount of memory it requires. I recommend that you take a look at this topic for more information about memory requirements for Exchange 2007 http://technet.microsoft.com/en-us/library/bb738124.aspx David.

Has anyone looked at this issue with Exchange 2007 and Windows 2003 SP2? http://msexchangeteam.com/archive/2007/07/18/446400.aspx

These could be a part of it. I have disabled chimney on my server and we will see how that works. I also changed my page file from a 6Gb to a 20Gb size and that is helping with all the memory being used up. I will keep you posted to the network issue. I updated my nic driver too but it only brought the ndis version to 5.2 so I thought it better to disable than upgrade the drivers alone.

Hello, I have been working on this issue since July of 2007. After talking with numerous people the following has appeared to help the most: Disable TOE on any Dell server and/or disable TCP Chimney and the following services: EnableRSS, andEnableTCPA. A level 2 tech at Microsoft told me that he has countless issues with the Dell 2950's and the broadcom networking cards and poor performance after a period of time with Exchange 2007. Hope this helps.

tadduci wrote: Hello, I have been working on this issue since July of 2007. After talking with numerous people the following has appeared to help the most: Disable TOE on any Dell server and/or disable TCP Chimney and the following services: EnableRSS, andEnableTCPA. A level 2 tech at Microsoft told me that he has countless issues with the Dell 2950's and the broadcom networking cards and poor performance after a period of time with Exchange 2007. Hope this helps. This FINALLY solved it for me. Exchange has been up for a week without anyreboots necessary, perfect backups every night. THANK YOU for this gem. Side Note: I did run into an issue with the Broadcom card. Run the disable chimney commands from the command prompt, do NOT just go into the software and try to disable TOE. My server wouldn't finish booting because it corrupted the network drivers and exchange would hang while starting services. Running the disable from the command prompt is the clean way to do this. This was a Dell Poweredge 2950 with a Broadcom Net Extreme II card. 1. Run: netsh interface ip set chimney disabled 2. Run: netsh interface tcp set global rss=disabled at the cmd prompt. 3. The above command may not work depending on SP2 and Revision so Try this registry key also: 1. Click Start, click Run, type regedit, and then click OK.2. Locate and then click the following registry subkey: HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters3. On the Edit menu, point to New, click DWORD Value, and then type EnableRSS.4. Double-click EnableRSS, type 0, and then click OK.5. Restart the computer on which you changed the EnableRSS value.

I think you are missing the point: Exchange is consuming memory to the point that Exchange fails. That's not memory management. You are hearing from users that have adequate configurations, and Exchange is failing. Where should we go for constructive help with this?

Can anyone suggest the recommended memory amount for my Exchange 2007 implementation?We have one serverwith 8 processor cores @ 2.66 GHZ per core with16 GB of memory. I have the Mailbox, Hub Transport and Client Access roles on this server. We are using LCR and SCR on this server too.200 usermailboxes are on the server.4 Storage Groups.The server is using 14.5 GB of memory. Store.exe is using 12 GB of memory. Should the server be consuming this much memory?

I like "features", but when the "feature" freaks out, then it's a problem. I got basically 100 users with 8GB and Store consumes around 4.5 - 5 Gb. Memory is at a constant 7.93GB or 93%. This is not normal. If it's a feature, is it just possible to disable the "feature" and give back the normal view. ;-)

Can anyone suggest the recommended memory amount for my Exchange 2007 implementation? We have one server with 8 processor cores @ 2.66 GHZ per core with 16 GB of memory. I have the Mailbox, Hub Transport and Client Access roles on this server. We are using LCR and SCR on this server too. 200 user mailboxes are on the server. 4 Storage Groups. The server is using 14.5 GB of memory. Store.exe is using 12 GB of memory. Should the server be consuming this much memory? Did you ever figure out what the problem was here? I've got a server that matches your specifications but I don't have the CAS role installed here...I split it out. So it's just a HUB/Transport and Mailbox role server. I also routinely see Store.exe RAM usage above 12GB.

Please view the following technet article regarding memory requirements for each Exchange role: http://technet.microsoft.com/en-us/library/bb738124(EXCHG.80).aspx As far as people wanting to decrease the memory consumption by the Information Store process, please remember that this is just caching doing its job. If you prevent the Information Store from growing in memory the way it does it may have a negative impact on the I/O performance of your servers, resulting in performance degradation and complaints from users. To put things in perspective my mailbox servers host 800 users each and have 32GB of RAM, and the store process consumes 28GB on average. The cache performance is fantastic, and I/O is greatly reduced from Exchange 2003. Store.exe will do its best to leave only 2GB of RAM, so having a system with 14GB of RAM means store.exe will typically consume 12GB. Taken from here: http://msexchangeteam.com/archive/2007/01/15/432199.aspx "To estimate database cache size subtract 2048MB (3072MB with LCR) from the total amount of memory installed in the Exchange Server, and divide that amount by the number of users."

Please view the following technet article regarding memory requirements for each Exchange role: http://technet.microsoft.com/en-us/library/bb738124(EXCHG.80).aspx As far as people wanting to decrease the memory consumption by the Information Store process, please remember that this is just caching doing its job. If you prevent the Information Store from growing in memory the way it does it may have a negative impact on the I/O performance of your servers, resulting in performance degradation and complaints from users. To put things in perspective my mailbox servers host 800 users each and have 32GB of RAM, and the store process consumes 28GB on average. The cache performance is fantastic, and I/O is greatly reduced from Exchange 2003. Store.exe will do its best to leave only 2GB of RAM, so having a system with 14GB of RAM means store.exe will typically consume 12GB. Taken from here: http://msexchangeteam.com/archive/2007/01/15/432199.aspx "To estimate database cache size subtract 2048MB (3072MB with LCR) from the total amount of memory installed in the Exchange Server, and divide that amount by the number of users." What you're saying is this is a feature... The problem is that store.exe crashes when it consumes too much memory...to the point of crashing every 2 weeks for a server with 8 CPUs and 16GB of RAM. To me, that's not a feature...it's a problem. When you combine roles (a la small business server) like HubTransport and Mailbox together on the same box...it really starts to cause major problems. That's where I'm at...I'm not on small biz server but I do have combined roles. Racing memory up to leaving 2GB leaves too small a window for error IMHO and obviously is causing instability in my case.

If you did not observe any other performance issues, it is normal that much available memory will be consumed by store.exe. A few articles on this, http://msexchangeteam.com/archive/2004/08/02/206012.aspx http://msexchangeteam.com/archive/2008/08/06/449484.aspxSajjad