Running Andy Fyfe's packdisk -- Any horror stories?

[John B. Milton]

In article <35887 at shemp.CS.UCLA.EDU> kirkaas at oahu.cs.ucla.edu (paul kirkaas) writes:
>
>My 3b1 hard disk has 46% fragmentation.  Time to do something about it,
>I guess.  I'm going to try and run Andy Fyfe's Packdisk program, but
>I've never done anything like it and am a bit leary.  Anyone out there
>have any experience with it?  Advice, cautions, bad stories?

It's a great program. It works. It has strange ideas about how to organize the
disk. It trys very hard to be extra careful, and thus is VERY VERY slow. In
every practicle test I did, it would have been significantly faster to do a
dump and restore to floppies. I started hacking on it, but haven't worked on it
lately.

A good way to test it is to let it play around with a floppy file system:

$ # put a scratch floppy in the drive
$ fdfmt10.nl
$ mount /dev/fp021 /mnt
$ cd /etc		# make a messy file system
$ cp * /mnt		# DEL when you start getting errors
$ rm /mnt/*[a-m]	# zap ~ 1/2 of the files
$ cp * /mnt		# overwrite/create, DEL when you get errors
$ umount /dev/fp021
$ fsck -y /dev/rfp021	# ALWAYS DO THIS BEFORE RUNNING PACKDISK!!
$ packdisk /dev/rfp021

I assume you got your fragmentation numbers from fsanalyze. I got fsanalyze to
a state where it seemed to be working, then did the logical thing: check a file
system before and after packdisk. According to fsanalyze, there wasn't much of
a change in the fragmentation. When you run packdisk a second time, it finishes
up right away, so it recognizes that the file system is still in the order it
likes. The output from fsanalyze doesn't seem very easy to use. Yet another
neato program I haven't had the time to dig into.

I also got it to work on this machine at work (celebr), which is a Unisys
6000/50 (a.k.a. Convergent Server/200). I don't think packdisk can really deal
with the bit map free list, so on this machine you MUST always followup with
an fsck -s.

I have worked out all the wrinkles with the HD2 now. I will be posting a list
of all interested buyers soon.

John
-- 
John Bly Milton IV, jbm at uncle.UUCP, n8emr!uncle!jbm at osu-cis.cis.ohio-state.edu
(614) h:252-8544, w:469-1990; N8KSN, AMPR: 44.70.0.52; Don't FLAME, inform!