Re: 1.3.73 goes ballistic

linuxsys@ssg.com
Wed, 13 Mar 1996 12:38:41 GMT


Thomas.Koenig@ciw.uni-karlsruhe.de writes:
> I wrote:
>
> >A short time after 1.3.73 went multiuser (named had just been started up),
> >the kernel went totally ballistic on me. There was a lot of what I could
> >vaguely see as a kernel oops or something related; however, it scrolled
> >by so fast that it was impossible to make out anything.
>

Alas, both .72 and .73 did the same here, too. .72 stayed alive for a
few hours and then began to spasm while trying to build .73. I can't
find any obvous error messages but /var/adm/messages began to fill
with CTRL-A's which is probably not normal behavior. <g> I was unable
to shut the system down with a "three-finger salute" or with "shutdown
-r now". I got as far as sending KILL but couldn't dismount the file
systems. When rebooting into .71 (which was up for 6+ days before
going to .72) fsck gave up on the wreckage and required a manual
filesystem check. /dev/sda1 was a mess with over two dozen inodes
shoved off to /lost+found. .73 lasted only a few minutes before
applications like emacs locked up with problems while trying to open
files (it took several *minutes* to open a file and yamm showed emacs
receiving SIGSUSPEND).

FWIW, this is a P100 with 32M and an ISA 1742 SCSI controller for the
root drive (/dev/sda1) as well an IDE drive.
linuxsys@wxsat:~:4> df
Filesystem 1024-blocks Used Available Capacity Mounted on
/dev/sda1 480625 367099 87876 81% /
/dev/sda2 641711 16640 590819 3% /usr/local
/dev/hda3 217975 68641 137702 33% /usr/local/scratch
/dev/hda1 51334 30854 20480 60% /dos

Gory details are available on request.

Rick