[9fans] Freeze

[9fans] Freeze

[baldwin]

I can make my terminal reliably freeze up if I run out of memory.
Usually, running upas/fs with a large mailbox does it.  I sometimes
get up to 1,000 mail messages, and my resulting 10 MB mailbox
blows up to 60-70 MB of memory used by upas/fs.  When that happens,
my terminal just completely freezes.  Mouse stops tracking.
No keyboard input.  Not even ^T^Tr works.  No "out of memory" prints
from the kernel on my console.  I do have swap enabled, which may
be a bad idea.

Re: [9fans] Freeze

[Boyd Roberts]

yeah, i had a suspicion that it was memory related.

rob, of course, will correct me ...

Re: [9fans] Freeze

[jmk]

On Wed Oct  3 20:53:20 EDT 2001, boyd@fr.inter.net wrote:
> > Yes. It's called trying to fit a quart into a pint bottle.
> 
> i believe it's called 'a failure of vision'.

absolutely. and when your vision improves i'm sure you'll
let us know.

Re: [9fans] Freeze

[jmk]

On Wed Oct  3 17:55:20 EDT 2001, boyd@fr.inter.net wrote:
> from time to time, when i run a meta-mk on python i get
> a freeze;  rio chokes and ^t^tr doesn't even work. fortunately kfs doesn't smashed itself, even though i have backups.
> 
> i _suspect_ it's a memory related problem.
> 
> is this a known problem?

Yes. It's called trying to fit a quart into a pint bottle.

Re: [9fans] Freeze

[Boyd Roberts]

> Yes. It's called trying to fit a quart into a pint bottle.

i believe it's called 'a failure of vision'.

Re: [9fans] freeze

[Martin Harriss]

Boyd Roberts wrote:
> 
> from time to time, when i run a meta-mk on python i get
> a freeze;  rio chokes and ^t^tr doesn't even work. fortunately kfs doesn't smashed itself, even though i have backups.
> 
> i _suspect_ it's a memory related problem.
> 
> is this a known problem?

FWIW, I've recently seen what happens when a terminal runs out of
memory.  (I was trying to use upas/fs on a large POP mailbox:)  The fs
process chews up all the memory, gets a trap and becomes broken (still
using memory.)  If I then try and run a command from a rio window, the
command fails (no memory to fork the new process.)  This appears to get
the system very upset, and it panics shortly thereafter.

(This did actually prompt me to clean up my mailbox, but I still don't
think I've got it small enough to make upas/fs happy.  Need to try IMAP
next.)

Martin

Re: [9fans] freeze

[Andrew Pochinsky]

FWIW, I had rio chocking on distributions after 5/17/01 on a P4 machine as
well as Thinkpad 570. It did not happen with older dists.

--andrew

   Boyd Roberts wrote:
   > 
   > from time to time, when i run a meta-mk on python i get
   > a freeze;  rio chokes and ^t^tr doesn't even work. fortunately kfs doesn't smashed itself, even though i have backups.
   > 
   > i _suspect_ it's a memory related problem.
   > 
   > is this a known problem?

Re: [9fans] freeze

[forsyth]

[-- Attachment #1: Type: text/plain, Size: 563 bytes --]

i've had thinkpads freeze.  i suspected the disk and/or bios power management interface.
i don't enable apm on the thinkpad but even tickling it in the initial load
seems to change the bios's ability (that's to say that the old system that
left it alone allowed the bios to do some things, i suppose with the smi
interrupt, and now it can't reliably do all that it did).

more specifically, the disk activity light is on, the (soft) power switch on
the thinkpad doesn't work, and i need to prod the little button hidden inside
to get the machine back.


[-- Attachment #2: Type: message/rfc822, Size: 2056 bytes --]

From: "Boyd Roberts" <boyd@fr.inter.net>
To: <9fans@cse.psu.edu>
Subject: [9fans] freeze
Date: Wed, 3 Oct 2001 23:50:58 +0200
Message-ID: <029401c14c55$7c6bb850$a2b9c6d4@SOMA>

from time to time, when i run a meta-mk on python i get
a freeze;  rio chokes and ^t^tr doesn't even work. fortunately kfs doesn't smashed itself, even though i have backups.

i _suspect_ it's a memory related problem.

is this a known problem?