|
PureBytes Links
Trading Reference Links
|
Let me make a suggestion - the next time any of you run Diskeeper and find
fragmented files remaining afterward, check the text display and note the
names of the fragmented files, especially the worst ones. Do any of them
have a pathname which points to a nonexistent subdirectory below the
Diskeeper directory itself? Or to the drive's root directory?
One of my three drives, an NTFS drive named E:, has four persistent
"fragmented files", all Omega files. What is very strange is that they are
ghost files - they existed once upon a time, but were deleted several weeks
ago and no longer exist on E: (they exist only as a backup on F:).
Searching the entire drive for these files finds nothing. I have searched
using DIR <filename> /A:H to see if by any chance they have been
reincarnated as hidden files. I have also searched using XDIR - no luck.
The pathname reported for the files varies - for a typical file it first
appeared as:
E:\dskpr\dklite.fts\40sig5.dat
There is a FILE named dklite.fts below E:\dskpr, but not a directory! If I
rename or move the file dklite.fts to another location, the ghost file is
then reported as:
\40sig5.dat
Last week when I ran Diskeeper (actually, the Lite version), the ghost files
had disappeared. But guess what - when I ran today, the files had
reappeared (in the dklite.fts "directory")!
Finally, after performing my usual weekly disk cleanup and defragging, I
deleted the file dklite.fts, rebooted, and reinitiated TS (TS was not
running when the above-reported results were obtained). I just reran
Diskeeper analysis, and NOW the files are reported as:
E:\omega\ts\ti006000.~da\<filename>
Again, there is a FILE named ti006000.~da, but no such directory!
My hypothesis is that somehow, Diskeeper is detecting a remaining fragment
or fragments of the four ghost files and, because they no longer have valid
entries in NT's file table, they appear with an accidental directory
pathname, perhaps whatever Diskeeper last looked at.
This is obviously a bug in Diskeeper and, although the symptoms I've seen so
far are relatively harmless, it is troubling nevertheless. Being a
programmer, I know that a bug which produces such illogical and varied
symptoms can also, if the conditions are right, cause a catastrophic
failure.
My recommendation: be on your guard, check the text display in Diskeeper,
and watch out for similar bizarre behavior.
Carroll Slemaker
|