[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: need help from fs guru



Hi,

On Fri, Feb 08, 2002 at 08:30:41AM -0500, Mohammad A. Haque wrote:
> On Fri, 8 Feb 2002, Cameron Simpson wrote:
> 
> > On 19:45 07 Feb 2002, Mohammad A. Haque <mhaque haque net> wrote:
> > | 
> > | 622669  40755      0      0    4096 13-Jan-2002 22:39 home3
> > | 2133571369 --- error ---  .autofsck
> > | 2133571369 --- error ---  websites
> > | 
> > | Does this mean there's a chance of recovery? If so, can anyone help? 
> > | It's not a great loss if it's not. I would like to recovery this is 
> > | possible since there are some items i'd like back that had not made it 
> > | to the backups before this happened.
> > 
> > Did it perchance relink it into the lost+found top level dir in that fs?
> > That's the normal place for unlinked files.
> 
> Update: I forgot /websites was a symlink so no big loss. Though, I'm 
> concerned that this is showing up in debugfs. .autofsck was recreated 
> (by the system i assume) and i recreated the symlink. The two errors 
> still show up in debugfs.

Running debugfs on a live filesystem is dangerous and can show
misleading information --- there is no cache coherency between
debugfs's buffer-cache view of the disk and the filesystem's internal
inode and directory caches.  If you've already populated the
buffer-cache view by running fsck, then it's entirely possible for
debugfs to be viewing that old copy even after the fs has modified the
inodes concerned, so that debugfs will see inconsistent data.  You
need to forcibly flush the partition's buffer cache to get around
that.

Cheers,
 Stephen





[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]