The preamble:
I have the following setup -
2 Servers consisting of Windows 2003 R2 on Dell Poweredge 2950, dual quad core Xeons, 8GB RAM, MD1000 external RAID unit
Both servers are setup with 5 network shares each (Shared, Groups, Users and some unused others).
Both servers are setup with the Windows Server 2003 R2 version of DFS to handle fail-over et al. The staging areas for these DFS enabled shares are on separate disks to the shares themselves.
Both servers have SQL Server 2005 Standard installed.
The problem:
I have two users out of 150 who have intermittent freezes when reading or writing files on the network shares - both are developers running on Windows Server 2008 systems as desktops (although their issues existed when they were on Windows XP Pro systems a month ago).
Both of these 'problem' users are developers.
No one else in the company has reported issues, despite it being specifically asked at various meetings and a straw pole being taken by myself around the company.
Both developers get the same slowdown at the same point in time, and they last for the same amount of time. Other users can access the same network shares during this period without issue.
Neither developer shares the same network path to the server, aside from the final two hops which everyone shares.
Neither file server is under load, both servers are typically at idle with plenty of RAM free. Since moving the staging areas off the same disks, both servers disk queues are as expected and not consistently at peak.
The question:
Can anyone suggest a way to trouble shoot this? I have a feeling that most of their issues are due to the development tools they are using (Slickedit), but I would like to eliminate all possibilities along the way.
I have monitored the servers using the various Sysinternals tools, which lead to the staging areas being moved, but aside from that there are no obvious issues.
Actually if it's only two users I'd start by looking at the physical layer. Does the switch or client NIC show any errored packets, flapping etc, is it possible the cable's dodgy, are they possibly bandwidth constrained?
Second if it's fairly easily reproducable I'd dump a packet sniffer on the machines (using a mirror port or transparent sniffer) and see if there's anything obvious at the protocol level.
Do they have two network cards? I've seen this when two network cards are installed, though in my case it seems to be related to the AVG anti-virus as the problem disappears if I uninstall AVG.
JR
Could be something to do with SMB 2.0 on the Server2k8 machines.
http://en.wikipedia.org/wiki/Server_Message_Block