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

Re: Corrupted disk labels

    Date: 22 Feb 90  3:13
    From: Dan Razzell <razzell@cs.ubc.ca>

	Date: Wed 21 Feb 90 15:24:26-PST
	From: Mabry Tyson <tyson@ai.sri.com>

	Just a reminder:  If you have more than one Symbolics that use the same
	disk types, you can often mount the problem disk as, say, the second
	disk on a running Symbolics.

	If the failed disk only has worlds that you can rebuild or copy from
	somewhere else, you may just want to reformat, (if you might turn off
	the machine or didn't SCAN the disk flods, you should have the
	breath-of-life tape handy), rebuild the initial file system (which
	specifies where the bad blocks are) from the tape for that disk drive,
	load a world onto the machine, and go on from there.

	If the disk had LMFS files that you can't easily recover, you probably
	do want to try to recover without reformatting by swapping the disk
	to another machine (by cross-connecting the cables and all the other
	details (like unit select numbers)).

	Date: Wed, 21 Feb 1990 17:36:00 UTC-0800
	From: Kalman Reti <reti@stony-brook.scrc.symbolics.com>

	If you need to recover files from a partially broken LMFS, I have a
	prototype tool that I'd be glad to supply.

    These are nice thoughts, but this is a standalone machine, and there is
    no alternate disk I might use to boot Lisp.  That's why I was so
    interested in knowing what could be done from the FEP.  The answer,
    apparently, is reformat and little else, so by the time I get Lisp going
    again, the old LMFS will be history.  It's a kind offer, though.
I think you can use the BOL tape to attempt to just write the initial fep
filesystem, without reformatting.  You can also only reformat the cylinders
that need it.  If you do that, you may still be able to recover LMFS files
that weren't on the cylinders formatted or of the files restored from the
BOL tape.

    The standard FEP flod files that I have on cart don't provide any real
    disk diagnostics, and the EMD368 drive doesn't have any status lights on
    it, so I have no idea if the disk is dead, acknowledging SMD commands,
    reporting faults or what.  I had hoped that someone would be able to
    point to some diagnostic flods, or even that the FEP Disk Format command
    had a nice little hammer in it for repairing bad labels, which I could
    at least try before resorting to more destructive measures.  
Doesn't the Test flod have a test disk command?  Unfortunately, I don't remember
exactly what it does and there's no one around to ask at the moment.  It certainly
reports the state of the label and whether or not it can read from various heads
and cylinders.
								 Sigh.  It's
    hard to believe, given an architecture with a separate Front End
    Processor, that nobody has thought of this before.
They've thought of it before, but the amount of memory and the horsepower of
the FEP is **EXTREMELY** limited, and the difficulty of writing FEP code is
*VERY* high, so many wished-for and dreamed-of features never got implemented.

You should be able to tell if the drive is at least working by typing SHOW DISK LABEL
and listening for head movements.  The FEP will retry reading each label block
several times before reporting too many retries.

If it doesn't report any such error (i.e. just hangs) the problem may well be with
the 36xx's interface to the disk rather than the disk.  If you are on maintenance
have your field servant check it out; if you are doing your own, make sure that
the I/O board an paddle are seated firmly, visually inspect the cables for cuts,
breaks, worn places, make sure the cables are seated firmly, and move the cables
near their connectors to different positions to see if that has any effect (one
favorite place for conductors to break is right where the ribbon cable meets the
connect, often invisibly).  If you have spare cables you could also try them.

In any case, there are several possibilities that I'd try before formatting the