low virutal memory stop

[homerunkevin]

Hi all

Hope you can point me in some good direction or just curious have you experience such scenario. At work, almost at least every day, I would need to restart Exchange Information System Store services. Why because the hard disk space will go from 2 GB to 744 MB left. When this happen, no one in the company can receive email, send email, or check into email server for any services. IT using so much virtual memory, it causes the email server to go down. In the beginning I thought, there were spam emails that cause the information store services or the SMTP to a hault. Sometime, in the mailroot/vsi/badmail/ folder there would lots of bad mail like at least 45MB of it. So, if you can help me a bit or gave me some hint, i would appreciate your help.

Love always,
Kevin Z
Techncial Support specailist

 

[bobbytupper]

homerunkevin - how much physical memory do you have in your mail server - how many users are you supporting - what volume of mail do you send/receive per day and how is your virtual memory configured?

 

[homerunkevin]

Bobytupper

HP LC2000 Server has 1GB of RAM, supporting 60 users, how much we send/receive per day, I'm not really sure. Can you tell how I can check that out. Virtual Memory is set to 1024-1500 maxium. Thanks for the tips.
Love always,
Kevin Z
Techncial Support specailist
kevin@hsmc.com

 

[Evilbart]

i suspect antivirus (i hope you have one ;-) )

see some other thread where i asked you the same

/Bart

 

[homerunkevin]

Bart,

Oh It's SYmentac AVF-antivirus filtering for Exchange.

THanks Love always,
Kevin Z
Techncial Support specailist
kevin@hsmc.com

 

[bobbytupper]

Homer - if you have 1gb memory with only 60 users I would not expect for the operating system to touch the page file. I trust this server does not run any other services that could be causing the problem (print etc.)

Do you get any error messages in the event log?

Is this the only exchange server in your organisation?

What service pack is installed?

See this White Paper from microsoft about store.exe and memory leakeage

http://support.microsoft.com/default.aspx?scid=kb;en-us;322125

Hope this helps.

 

[bobbytupper]

Ah one more thing you can monitor the volume of smtp traffic possibly on your firewall.

 

[homerunkevin]

Bobby,

Well, it's been upgrade to SP3? Not necessarily any specific error in the Event Viewer. Yeap this is the only Exchange Server in our organization.

Since Exchange Server was install on C:, i heard that it is not good, when the log file grow, where it will used the C: drive space. However, there is a partition D drive on the server that holds the storage group, database, .edb, .smp. file.

HOw do I monitor traffic for SMTP mail on CHECKPOINT? That is what we are running? Help!! =)

Thanks also for your time Love always,
Kevin Z
Techncial Support specailist
kevin@hsmc.com

 

[bobbytupper]

The way you are supposed to configure your drives is as follows

Operating system and swap file. fixed disc or mirrored set to achieve fault tolerance.

Exchange binary files and server application files. fixed disc.

Exchange database files. fixed disc or Raid to achieve fault tolerance.

Exchange transaction log files. fixed disc or mirrored set to achieve fault tolerance.

This is looks a little excessive I know but this is what happens incoming mail is first written into the memory (hence 1 gb of RAM ) then the mail is written into the transaction logs once it is successfully written into the transaction log it is then written from memory into the database. Once the transaction is committed into the database it is removed from memory. This makes exchange VERY disc intensive and if the cpu cant write from memory to disc fast enough due to lack of resources/disc time a back log effect will occur causing potential problems. I am presuming that with the size of the 4 SCSI there are plans for this server to be the office file server also which will mean a potential 100 user IO during working hours. Therefore if we put the transaction logs and/or on the same set of physical discs we could put too much load on the discs.

I am presuming that Manchester will be using the standard version of exchange which as we found out has a limit of 16GB. In this case we could easily fit the database onto 2 x 18GB drives mirrored if maxed. This of course would not allow for expansion (2x 36 recommended)

I would put the exchange binary files on the same drive as the OS (again presuming the 2x18gb drives mirrored as quoted)

This just leaves us with the log files. Now the log files are cleared (deleted) after each full backup. Therefore these drives wont total a great deal in size. We could either put the log files on say 1x18gb SCSI no fault tolerance or 2x18gb mirrored SCSI with fault tolerance.

This is probably a bit excessive in you case but if you arent taking regular back ups and the log files are on the same partition as your page file then maybe the log files are building up?

As far as monitoring smtp throughput, how is your firewall configured,

Do you use a one to one NAT configuration

Do you use a static NAT ( port Forward )

Does the Firewall proxy SMTP traffic?

You should be able to find documentation on the vendors website on how to graph the traffic.

If this fails you could always go for maybe snmp and graph using mrtg which is quite effective.