Sbs2K8 Mail Server Sporadically Loses Connection To Workstations

[dcestrada]

Thanks for taking a look! I would appreciate any help you can provide with this problem:

We're experiencing the following issue in a small office environment: several times an hour, workstations will lose connectivity to at least the mail server, causing an error message in Outlook (connection to the Exchange server lost) and causing Internet and intranet connectivity to timeout. Sometimes the connection is restored in a matter of seconds, and sometimes it's down for 10-15 minutes. The network connection on the workstation does not show any connectivity problems during this time, but you can't access Internet or intranet pages most of the time. The workstations do not all drop at the same time -- the outages are sporadic. A few of the workstations are laptops, and we've noticed that the problem mostly (or completely) goes away when we run those laptops on the wireless network connection instead of via network cable.

 

The setup is as follows:

- 7x Windows 7 workstations (4 laptops, 3 desktops) with Outlook 2007

- Linksys SR2024 switch

- Cisco WRSV4400n (v2) Router/wireless access point

- 2x Dell PowerEdge 2950 servers:

- Mail Server: Exchange 2007 SP3, SharePoint WSS 3.0, and DNS/DHCP on Windows Small Business Server 2008 Standard (64-bit)

- File Server: file & print sharing on Windows Server 2008 Standard (64-bit)

 

The issue appears to only affect the connections between the SBS2K8 mail server and the workstations, since shared network drives seem to stay online (hosted on the File Server). This issue just started a few days ago after the servers were moved into the same telco closet as the router & switch, instead of being connected through a jack in the wall in another room. The office has never had this issue before.

 

Here are the things we've tried ("server" below = mail server):

- Checked event logs on the server, nothing seems related nor consistent enough with this problem

- Checked cabling, there don't appear to be any issues with loops, etc.

- Set DNS server IP statically on a few of the workstations

- Flushed the DNS cache on the server and a few workstations

- Reset network connections on a few workstations using "netsh int ip reset"

- Changed settings on fileserver NIC (disabled flow control, ethernet wirespeed, interrupt moderation, etc.)

- Disabled RSS on the server NIC using:

netsh interface tcp set global rss=disabled

netsh interface tcp set global autotuninglevel=disabled

- Checked the ARP tables to make sure there aren't any duplicate MAC addresses or IPs

- Uninstalled Symantec Endpoint Protection client on server (management console still installed)

- Uninstalled Norton A/V software on a few of the workstations

- Rebooted server after making some of these changes

 

The closest article I've been able to find concerning this issue is here, and we've tried most of the suggestions outlined: http://social.technet.microsoft.com/Forums/en-US/winserverNIS/thread/9da2abd2-18e7-4ab5-babc-a98a1f7e4a15

Other options might include having the router manage DNS/DHCP and disabling that on the server, but I'm not sure that would make a difference (and DHCP is completely turned off on the router right now). We also have not tried uninstalling the NICs on the workstations or server, but that seems like an unlikely fix given that the workstations do not all disconnect at the same time. Also have not tried removing and re-adding the workstations to the domain.

 

Any other ideas? We're really stumped.

[ICTCity]

I'm quite sure your problem is your DNS server...

 

According to Microsoft, DNS should live alone. Exchange opens many connection, DNS does the same. I don't know if this is related to the movement of the server, basically it is plugged on the same switch...

 

Check your switch for any "flow control" or other protection system, and please consider to move your DNS on a standalone server or at least not the same of exchange.

 

Check the sysinternals website to find network utility which can help you to determine if your server has no more ports available.

 

Let me know.