At wits end on this issue.
Customer has a small network of 6 computers. 2x windows 7, 4x windows XP, 1x virtualized Windows NT4.0 Server.
One of the windows 7 machines is configured to act as a file server for the rest of the network,
It has 4 different shares containing around 2.5TB of files. It also has VMware workstation installed and hosts a virtual server running legacy application on Windows Server NT 4.0. There is no domain however, all computers in a workgroup and access the legacy application hosted on the NT server virtual.
All shares are mapped as network drives on the other computers.
Consistently having issues making the windows XP machines stay connected to the Windows 7 file server with their mapped drives. Its making me batty.
First, it was something which was happening over night only, works all day, but left over night and in the morning all the windows XP machines would report "Local device name already in use" error message. Meanwhile, the Windows 7 machines still access the shares no problem.
Restarting the XP machines wouldnt fix the issue either. They would still report local device name already in use after a restart. Only rebooting the Windows 7 machine hosting the shares would resolve the issue, and that meant that the VM would have to get shut on and off all the time which we don't like at all, the application database it hosts is old and seems fragile.
My first step was to look at how the sharing had been implemented. I turned off password protected sharing when first setup, so all computers were just connecting to the shares without a username and password. I though this might be a problem so I enabled password protected sharing and created an unique login account for each network device on the server.
So each windows XP machine now maps the network drives with its own unique username and password such as "desktop1" "desktop2" ETC. This seemed to calm things down, the XP machines stayed connected for around 2 or 3 weeks without dropping off until all of a sudden it started happening a lot, not just during the night, but throughout the day. One of the XP machines reported a unique error that the others had not. It was saying the cause of the issue was resource shortage, rather then the name in use error the others had. I discovered some log entries in the event log on the Windows 7 host regarding IRP Stack size. We use Acronis True Image, and I have stumbled accross issues in the past were this applications presence has messed with the IRPStack and caused these sort of problems.
Eureka, I figured that possible the issue was allways regarding the IRP stack, and that even the windows XP machines reporting device name in use were doing so as a result of this IRP stack size issue on server side, because device name in use is normally resolved by restarting the XP machine in my experience, though in this situation only restarting the server resolves it.
I increased the IRP stack to 25 on the Windows 7 host machine rebooted and the resource error was gone and this calmed down again for a few days until now, when the overnight drop outs started up again. Driving me nuts, before we implemented the Windows 7 machine to host all the shares we had a kick about cheap Thecus NAS that was slow as buggery but it worked fine, all these windows XP machines had no issues with mapping drives on the Thecus nas. I suggested we chuck it all on Windows 7 Computer as it was gigabit and would perform alot faster, they said great and now we have these annoying problems.
Any help in the right direction? Further increasing IRP stack size?? Event logs are clean for the IRP issue now mind you. There is however, a continuous error in the log every few seconds in relation to the lanmanserver. Only seems to have surfaced since increasing the IRPStackSize
Event ID: 2017
Task Category: None
The server was unable to allocate from the system nonpaged pool because the server reached the configured limit for nonpaged pool allocations.