On Thu, Feb 14, 2002 at 09:56:45AM -0800, Dan Wilder wrote:
> > 1:  atime gets touched.  Most of the time you don't care, but when you
> >     really need it (usually forensics for me), atime is nice to have.
> 
> Assuming atime isn't actually _used_ by amanda (I can't imagine it
> would be), you could hack the amanda source to add
> 
>   --atime-preserve

But then ctime moves.  TANSTAAFL.  :-)

> If this is a _big_ problem, the source for tar could be modified.

Usually not a big problem, and I've got around it in the past by spanking
the silly users that caused it... 

You would want to think *real* hard before just yanking the filename
and path length limits, because then a "normal" tar mightn't be able
to read it any more, and disaster recovery could become interesting.

Really, the two points above aren't huge, but just point out the
advantage of working from within the file system (ufsdump) instead of
through the file system if you can.

tar has many advantages of it's own, of course, not least that it
actually works well, today, on reiserfs/jfs/xfs/ext3 Linux file systems
for which no dump even exists (AFAIK), on ext2 file systems where dump
apparently does not work well currently, and that it is (mostly)
portable across different OS platforms.  *That* is the advantage of
working through the file system.

-- 
Jay Lessert                               [EMAIL PROTECTED]
Accelerant Networks Inc.                       (voice)1.503.439.3461
Beaverton OR, USA                                (fax)1.503.466-9472

Reply via email to