Nfs – Mount NFS on Windows XP

Excellent question I ran into the same problem some time ago. Windows is attempting to restore the mapped network drives before they unix services are ready. You can get rid of the error message but it will show up as a disconnected network drive until you double click on it. This has been tested on Windows XP SP3 Pro.

You can create a new dword in regedit

HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\NetworkProvider\

add a new key DWORD named

RestoreConnection

set it to 0

This will suppress the message as the drive won't attempt to reconnect before the unix services have been loaded. So far I haven't been able to find anything out about the "firing order" but this does remove the message so the system will continue to load. For me this was functional enough to move on to the other projects.

I did a quick tutorial on setting up NFS clients as I couldn't find anything about this online. http://www.hilands.com/os-windows-nfsclient.html

Usually when there are disk error like you are seeing, the disk tends to pause for a moment to try and correct the error itself, and the Linux RAID will tolerate some disk waiting until it marks is as bad. This disk pausing could be what is causing the slow downs (especially at the error rate you are seeing).

Your plan to replace the drive is correct. However, I wouldn't recommend getting a bigger driver with the notion of partitioning part for the replacement RAID and then part for use of something else later on. It would be more wise to just get closer to the original disk (size and speed) to keep the array consistent. That said, you COULD, in theory, go larger and partition the exact replacement size for the array replacement and then another partition for something else (even another member of an array).

A side note that may help with debugging: A tool I like to use as a replacement to top is called atop (http://www.atoptool.nl/) and it may give you more visualization into each disk and processes using disk I/O and where the bottleneck is (you may notice that the wait I/O is for the specific disk having the problem).