Hi all, I'm currently investigating issues with an Exchange 2007 Enterprise SP1 RU1 CCR mailbox cluster running on 2003 R2 SP2 x64 Enterprise with 16GB of RAM. Ive also applied PSP7.9.1, disabled the SNP and applied patches KB940307, KB944971, KB948354. I apologise for the long problem description, but this is driving me nuts at the moment. The problem, from what I can see, is when the server is under maximum load there will be an event trigger somewhere that causes upwards of 5-8GBofPrivate Byte memory allocation be stripped from the store process. Shortly afterwards the storage groups dismount themselves to keep the server alive as the transactions logs back up past the threshold forExchange 2007 (5120 logs per storage group).It seems to happen after the RAM allocation drops off leading me to think its the loss of RAM that causes thebuild up of transaction logs not the other way around. Timing around the issues is difficult to confirm as the logs don't play incompletely for hours after the load has ceased and the replication restarted. I have run perfmonon the primary node and logged the Private Byte allocation for all processes and also the Working Set for the store process. During the last failure the working set dropped in parallel to the Private Byte allocation, which also doesn't seem right... as I believe the store cache should maintain its hold on the RAM even if the pressure from the Working Set drops off, unless there is pressure coming from somewhere else or the store processshould page out the dropped allocation to allow the application putting the pressure on access to RAM. No other process increases its Private Byte allocation while all this is going on and I would expect if one did it would be easily visible in perfmon. Also interesting is that the page file size and utilisation does not change at all during this whole process (4GB page file, 7%-8% peakusage). This is a dev server attempting to replicate the issues on the production server, the prod server is running a 16GB page file, same issue. So, can anyone offer any ideas where I should be looking for the cause of this issue? The load on the servers is not beyond their capacity and I feel there is something going wrong with the memory management side of things, but I'm stuck as to where to find the cause or confirm it is an issue worthy of escalation. Thanks, Greg

Hi, Please check application log via eventvwr to see any related event logged there and then post here for further troubleshooting. Besides, you may refer to the below article: The Store.exe process uses almost 100 percent of CPU resources, and the size of the public folder store increases quickly in Exchange Server 2007 http://support.microsoft.com/kb/925252 Best regards, Xiu