Xref: utzoo unix-pc.general:4436 comp.sys.att:8387
Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!tut.cis.ohio-state.edu!snorkelwacker!spdcc!gst
From: gst@spdcc.COM (Gary Trujillo)
Newsgroups: unix-pc.general,comp.sys.att
Subject: Re: possible head crash  :-(
Message-ID: <1134@ursa-major.SPDCC.COM>
Date: 5 Jan 90 08:57:33 GMT
References: <1108@ursa-major.SPDCC.COM>
Reply-To: gst@ursa-major.spdcc.COM (Gary S. Trujillo)
Organization: S.P. Dyer Computer Consulting, Cambridge MA
Lines: 102

In article <1108@ursa-major.SPDCC.COM> gst@ursa-major.spdcc.COM
	(Gary Trujillo [me]) writes:
> Well, it may have finally happened...
> [details of possible head crash deleted]

Here's an update on the situation I described recently.

Sorry to burden you with my continued saga of woe, but (heavens
forfend) something like it just might happen to *you* someday,
so you may want to pay attention. :-)  I seem to be able to
access a couple of other machines where I can get at mail and
news using the 7300 I haven't touched in a couple of years.  I
think it would be OK to post any suggestions as followups to my
articles, as it appears I'll be able to read the UNIXpc news groups
reliably via this method.

Fortunately, I have backups of the important stuff that is recent
enough that there's not a lot I'd need to get off the disk in case
it turns out to be unsalvagable.

Well, I thought I might be able to get stuff off the disk by creating
compressed tar or cpio archives in /tmp (where I have > 10K blocks),
and pulling it over to the other machine with kermit.  The problem
turns out to be that I can't write to the disk, it seems.  I've been
able to rename a file, which requires a write to a directory block.
(I did a sync afterwards, just to make sure it was really written,
and not just cached for write.) However, I can't even cp a file: I
hear sounds of attempted recals, a number of clicks, indicating write
attempts, I assume, and then a message from the cp command saying the
copy failed.

So, my new questions are as follows:

1. What's down there at block 2, which fsck reports it gets a read
   error on?  I would have assumed that the bootstrap loader (or
   whatever it's called) lives there.

2. What are the chances I might be able to re-write the boot loader
   with the "ldrcpy" utility, which is in the /etc directory on the
   "floppy file system disk" (3 of 12)?  There are some lines in
   /etc/profile  on this floppy which say:

	# Copy the loader from the floppy to the hard disk
	echo "\nCopying the loader onto the hard disk ....\n"
	ldrcpy /dev/rfp020 /dev/rfp000

   I have my doubts that I'd be able to write *anything* to the boot
   sector of the hard disk, but I'm wondering whether ldrcpy knows
   how much to write, or it assumes the next thing you're going to
   do is an /etc/mkfs (since that's the next thing that happens in the
   /etc/profile script), so it can write as much as it wants, and
   needn't worry about smashing the superblock of /dev/fp002.

   (I tried "ldrcpy /dev/rfp020 tmpfile" - but it seems to require
    a special-device name as its second argument, so I can't seem
    to fake it out to see how much it wants to write.)

3. It turns out the file system is *not* completely OK.  I just ran
   an "ls -R /" for the bittersweet fun of it, and found there were
   several files (~ 10) which were reported "not found," amid the
   buzzing of recal attempts.  I assume the inodes of these files
   are inaccessible due to the damage which I now feel justified
   in imagining has taken place.

4. Speaking of damage, I suddenly realized that I might be wise to
   not run the unit too long, since, if there was a head crash, there's
   probably oxide flying around inside, which is not good on heads
   which are designed to float really, really close to the surface of
   the disk platters.

5. Here's a puzzler for you all-- given the conditions I've described,
   how can I get files off the machine? Here's what I have:

   A. A mostly-readable hard drive which I can't run a multi-user
      system from at present; the only way I can run UNIX is from
      the (writable) floppy filesystem disk I made quite some time
      ago in anticipation of major problems like the present ones.
      I cannot boot from the hard drive even starting from the boot
      floppy (I tried).  I can't seem to write to the drive.

   B. A 7300 with a 10meg drive (mostly filled, but I can probably
      get ~4K blocks free if I work at it).

6. Can anyone recommend a good replacement unit, if it comes to that?
   I have the WD2010, so I can run a drive with (what is it, John
   Milton? - 1400 cylinders?).  I don't have the P5.1 upgrade, but
   I'm sure I could get it easily enough.

7. Is there any point in attempting to reformat the drive, once I've
   gotten as much as I can from it, do you think, or would you just
   leave it in its current state.  (I know it's really up to me, but
   I'd sort of like to know what you'd do and why.)

8. What do you guess are my chances of saving the drive?

Well, that's enough questions for one night!  If anyone wants to
send answers via email, I'll summarize what I get.  Otherwise, please
post your replies.

Thanks to everyone!  (And I hope your new years are happier than mine.)

	Gary