Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!columbia!rutgers!ucla-cs!ames!oliveb!pyramid!voder!kontron!cramer
From: cramer@kontron.UUCP (Clayton Cramer)
Newsgroups: comp.sys.ibm.pc
Subject: Re: Lost clusters, chains, and files
Message-ID: <1637@kontron.UUCP>
Date: Wed, 10-Jun-87 14:06:47 EDT
Article-I.D.: kontron.1637
Posted: Wed Jun 10 14:06:47 1987
Date-Received: Sat, 20-Jun-87 08:45:39 EDT
References: <1855@isis.UUCP>
Organization: Kontron Electronics, Mt. View, CA
Lines: 36

> Earlier this evening, when running a "chkdsk b:*.*", I got the
> wonderful and mysterious message:
>    1 lost cluster found in 1 chain -- 
>     Convert lost chains to files?
> 
> What does this mean? The chkdsk went on to say that the disk itself
> had all sorts of file system errors .. Can one recover from this 
> "lost cluster condition" without reformatting the disk?
> 
> Many thanks..
> 
> Ted Koppel == CARL - Colorado Alliance of Research Libraries ==

"Lost cluster" means that a cluster of sectors were allocated in the 
FAT (file allocation table), but not attached to any file in the 
directory.  This is frequently the result of a program writing a
file, then the system crashes before the file was closed.  The directory
is therefore out of date, but the FAT shows the cluster as allocated.

Use the /F option on CHKDSK.  This will ask you if you want to free
the lost clusters, or turn them into files.  If you turn them into
files, they go in the root directory with names like "FILExxxx.CHK",
and contain the contents of these clusters.  I've never had any real
use for trying to reconstruct files from these lost clusters.

Before anyone flames PC-DOS for not keeping the directory up to date,
remember that there is a trade-off between perfectly up to date file
system, and performance.  I wrote a file system for a telephone switch
several years ago where the old file system trade-off had been on the 
side of performance -- the wrong choice for a system that had to come 
up automatically and lose no data.  For a system like PC-DOS, Microsoft
chose performance, and I can't claim that this was automatically a
mistake.  (Sure would be nice to have a flag you could set to make
the trade-off at run-time...)

Clayton E. Cramer