To start off I have seen several posts that touch on a similar issue but none of them reach a conclusive solution so it is time for me to ask for help directly.
Setup:
2 NLB Front End Exchange Servers / Windows Server 2003 SP1 / Exchange 2003 Std
4 Backend Exchange Servers / Windows Server 2003 SP1 / Exchange 2003 Enterprise
For naming sake lets call the servers the following so you can see the individual hardware specs.
Front End Servers
exfe1.domain.internal - Dual Xeon 3Ghz Dual Core / 4GB RAM / Mirrored 15K 76GB HD
- no mailboxes
- primary SMTP relay server for company
exfe2.domain.internal - Dual Xeon 3Ghz Dual Core / 2 GB RAM / Mirrored 15K 76GB HD
- no mailboxes
- used as "internal" host for OWA since NLB doesnt play nice with local Entourage clients
Back End Servers
exbe0.domain.internal - Dual Xeon 3Ghz / 3GB RAM / 8 136GB HDs in 1 Virtual disk partitioned into a "c" and "e" (I understand this is not optimal and am in the process of phasing this server out to rebuild and reuse with 2 virtual disks)
- hosts approx 50 power email users mailboxes (average mailbox is approx 1GB)
- does not host a public folder store
- transaction logs are on the c drive
- exchange db is on the e drive
- smtp queue is on e drive but is relayed to exfe1.domain.internal so it does not store anything
- performance is good on this server with slight exceptions (below)
- 1 Entourage client conects to this server
exbe1.domain.internal - Dual Xeon 3Ghz / 4GB RAM / Mirrored 15K 76GB virtual disk "c", 4 - 136GB 15K drives in RAID 5 virtual disk "e"
- hosts approx 70 power email users mailboxes (average mailbox is 1 GB)
- hosts public folder store
- is the HQ server, most clients that access this server are in the same building as this server
- transaction logs are on the c drive
- exchange db is on the e drive
- smtp queue is on e drive but is relayed to exfe1.domain.internal so it does not store anything
- MAJOR DISK QUEUE ISSUES ON THIS SERVER
- 30 Entourage clients connect to this server
exbe2.domain.internal - Dual Xeon 3Ghz / 4GB RAM / Mirrored 15K 76GB virtual disk "c", 4 - 136GB 15K drives in RAID 5 virtual disk "e"
- hosts approx 800 email users mailboxes (average mailbox is 100MB)
- hosts public folder store
- is the remote office email server, most clients that access this server are not in the same office
- transaction logs are on the c drive
- exchange db is on the e drive
- smtp queue is on e drive but is relayed to exfe1.domain.internal so it does not store anything
- performance is good on this server
- all client access to this server is via Outlook 2003 using RPC over HTTPS or OWA, there are no Mac clients connecting to this server
Client machines are either:
Power Mac G5's (non intel) running OS X 10.4.9 with Entourage 10.4.4
Windows XP Pro SP2 with Outlook 2003 SP2
Before I go way deep into what is wrong I will tell you what has already been done.
- All hardware has been tested throughly
- All servers were stress tested before entering production with no issues
- The Exchange Best Practices has been run and reports NO issues
- Connectivity to AD and domain trust has been tested throughly with no issues
- ALL UPDATES/PATCHES/SERVICE PACKS have been applied for client applications (entourage and outlook) and client OS's (OS X and Windows XP Pro) and servers EXCEPT for SP2 for Windows Server 2003
The only problem I have relates to "exbe1" and the Entourage users's mailboxes. The server reports an ESE 507 error about twice a day in the application log. This error reflects a significant lag in disk access on the main EDB file (the one that hosts the Entourage mailboxes). These mailboxes have been moved twice in the past year and the issues that I have seen have followed them. I am positive that this server is capable of handling these mailboxes.
I have monitored the machine with sysmon and the following counters
- physical disk/% Disk Time/ - should be less than 50 but at times can spike to 80+
- physical disk/Avg. Disk Write Queue - Should be less than the amount of spindles in the virtual disk (in this case that is 4), but spikes to well over 200 at times, it can also return to under 1 after Entourage users have shut down their client.
Does anyone know what I can do to better monitor the Entourage clients. Exmon only logs MAPI access and I am confident that these issues stem from Entourage 2004 and WebDav.
I am not sure what else I can tell you so I will await for replies and hope that we get to the bottom of it.
Thanks,
Alex
Start Free Trial