Return to BSD News archive
Path: sserve!manuel!munnari.oz.au!sgiblab!sdd.hp.com!uakari.primate.wisc.edu!ames!bionet!raven.alaska.edu!sxjcb.uacn.alaska.edu!user
From: sxjcb@orca.alaska.edu (Jay C. Beavers)
Newsgroups: comp.unix.bsd
Subject: Crashing the System
Message-ID: <sxjcb-310792134054@sxjcb.uacn.alaska.edu>
Date: 31 Jul 92 21:48:35 GMT
Sender: news@raven.alaska.edu (USENET News System)
Followup-To: comp.unix.bsd
Organization: University of Alaska Computer Network
Lines: 24
Nntp-Posting-Host: sxjcb.uacn.alaska.edu
I've found 386BSD to be a fairly robust system so far, but twice yesterday
I managed to crash the system hard (no recovery -- once it tried to save to
disk and reached the ramdisk part before it totally died, once it just
froze solid) by doing pretty large tasks.
The first time I was doing two cpio -it's simultaneously with the input
coming from one NFS server and the output going to another NFS server on
the src01 dist and the etc01 dist. The swap daemon was burning some pretty
good CPU time and these tasks went on for ~45 minutes before the system
crashed.
The second time I was compiling libc and after the system crunched for ~45
minutes once again the system crashed solid.
My system is a 386/25 with 4 mb ram and <10 mb swap space on disk, so I
suspect that out of memory errors caused the crashes, but why was there no
warning at all? Is there no memory diagnostics in 0.1 and is this
something I should pay special attention to avoid?
______________________________________________________________________________
| jay@seaspray.uacn.alaska.edu
Jay C. Beavers | sxjcb@orca.alaska.edu
University of Alaska Computer Network | sxjcb@alaska.bitnet
________________________________________|_____________________________________