From Moon at STONY-BROOK.SCRC.Symbolics.COM  Fri Jul 29 17:22:00 1988
From: Moon at STONY-BROOK.SCRC.Symbolics.COM (David A. Moon)
Date: Jul 29 88 11:22 EDT
Subject: MX falls over, tapes flutter
In-Reply-To: <419832.880728.CENT@AI.AI.MIT.EDU>
Message-ID: <19880729152207.6.MOON@EUPHRATES.SCRC.Symbolics.COM>

    Date: Thu, 28 Jul 88 23:48:02 EDT
    From: "Pandora B. Berman" <CENT at AI.AI.MIT.EDU>

    ....
    a possibly related problem: the second time this happened, i was running
    into a tape flap error. what happens -- apparently reproducibly with AI
    tape 1088 -- is that while reading the tape, it hits some region about 3/4
    of the way in that it doesn't like. the vacuum columns start to flutter.
    the tape continues feeding into them, slowly, until it apparently reaches a
    max enforced by the hardware. then the tape in the columns just flutters,
    not very far up or down, but a lot; the tape on the input spindle shakes in
    concert with them, but i couldn't tell whether the takeup reel was moving
    at all. the scary thing is that this aberrant behaviour produces no err
    msgs, so i can't tell what to do.

This behavior is what the drive looks like when there is a read error and
the software keeps retrying the read.  The tape is actually going forwards and
backwards about 9 inches very quickly.  I'm surprised it doesn't eventually
give up and report a hard read error.  Maybe some bug.  You might end up not
being to copy that tape.

From CENT at AI.AI.MIT.EDU  Fri Jul 29 05:48:02 1988
From: CENT at AI.AI.MIT.EDU (Pandora B. Berman)
Date: Jul 28 88 23:48:02 EDT
Subject: MX falls over, tapes flutter
Message-ID: <419832.880728.CENT@AI.AI.MIT.EDU>

tonight MX got 2 errors of the form
 UUO while in AC BLK 0
 BUGHALT....
within about 2 hours. alan looked at the first one and said "but it can't
do that!" and explained that that purported to convey that the SYS job was
running with accumulator block 0 selected, which i gather is a bad thing.
when the second one occured he was on his way out the door, and suggested
that if it keeps getting UUOs like this, maybe the processor is actually
broken and needs to by fixed by (gulp) DEC. i have not touched it; someone
competent please investigate.

a possibly related problem: the second time this happened, i was running
into a tape flap error. what happens -- apparently reproducibly with AI
tape 1088 -- is that while reading the tape, it hits some region about 3/4
of the way in that it doesn't like. the vacuum columns start to flutter.
the tape continues feeding into them, slowly, until it apparently reaches a
max enforced by the hardware. then the tape in the columns just flutters,
not very far up or down, but a lot; the tape on the input spindle shakes in
concert with them, but i couldn't tell whether the takeup reel was moving
at all. the scary thing is that this aberrant behaviour produces no err
msgs, so i can't tell what to do.

NB: over the past several days the front-end 11 has crashed twice, causing
the sysconsole to go west. the first time, this weekend, i used a bigger
hammer by rebooting the whole machine. when it happened again earlier this
evening, alan tried to apply sweet reason to it, but it didn't work, so
he had to boot MX. but that wasn't what caused the lack of err msgs about
the tape flutter, because the sysconsole was entirely capable of printing
the UUO err msg.

see the system log. no film at 11.

From CENT at AI.AI.MIT.EDU  Tue Jul 26 08:16:06 1988
From: CENT at AI.AI.MIT.EDU (Pandora B. Berman)
Date: Jul 26 88 02:16:06 EDT
Subject: poor old kl hardware sux rocks
Message-ID: <418290.880726.CENT@AI.AI.MIT.EDU>

it's been crashing on me every 10-15 minutes for the past 2 hours.
probably i have been forcing it to over-exertion by trying to read a
remote tape -and- (via supdup) to catch up with my old mail at the
same time. the remote tape is a copied KA full, which i am trying to
list to a file; i suppose i'll just have to do that elsewhere while
using the KL to send bits out.
     i don't mind the constant too many par err problems -- at least,
i have been managing to make progress despite them. but now unit 0 has
its UNSAFE light on. i don't want to deal with that at this hour.
someone else please take a look at it. thanks.

From CENT at AI.AI.MIT.EDU  Sat Jul 23 08:17:38 1988
From: CENT at AI.AI.MIT.EDU (Pandora B. Berman)
Date: Jul 23 88 02:17:38 EDT
Subject: patched MX system
Message-ID: <417298.880723.CENT@AI.AI.MIT.EDU>

thanks to moon, MX is running MTPITS, a version of NITS patched to be
especially nice to tape software. it should continue running MTPITS
until the tape copying project is done.

From DEVON at AI.AI.MIT.EDU  Sat Jul  2 20:04:15 1988
From: DEVON at AI.AI.MIT.EDU (Devon Sean McCullough)
Date: Sat,  2 Jul 88 14:04:15 EDT
Subject: No subject
Message-ID: <406657.880702.DEVON@AI.AI.MIT.EDU>

I found I had a detached tree from yesterday (phone lines here are ATROCIOUS) and it had an emacs in it...so I logged in and did alt-G at the emacs and I got --Undefined Symbols-- which to me says there's been bitrot.