>>>Oh phooey, it's Friday isn't it.:-) I didn't mean it to be taken so literally. Whatever file system is used (same results on both NT and Novell). Darn acronyms:-).
>>
>>I don't think it really matters. If the drive controller is caching the write, then the OS may not have detected the change yet.
>
>It's still Friday over here (and it's been a looooong one), but why should the controller's caching matter at all? If the info is not physically written yet, it doesn't matter - any decent controller would immediately return the fresh info from the cache, not moving any of its heads (it has seven heads, but it's not a dragon - it's a disk). I guess that's also quicker to retrieve it from the cache - wasn't it the purpose of the cache in the first place?
>
>OTOH, I wouldn't trust a disk which doesn't read its own (write or any)cache.

This is a common problem with databases...even high end disk controllers do things this way.