swap on disk1

Norman Kohn nvk at ddsw1.MCS.COM
Mon Jul 23 14:28:02 AEST 1990


In article <1990Jul17.183807.6417 at hawkmoon.MN.ORG> det at hawkmoon.MN.ORG (Derek E. Terveer) writes:
>Speaking of swapping --- i have been experiencing some problems with adding swap
>areas (with the "swap -a" command) on my ESIX 5.3.2-C system (i experienced the
>same problems on my microport system as well).  Sometime soon after adding a
>swap area on a second disk (either scsi or mfm), i get a panic message along
>the lines of "i/o error in swap" and the system crashes.

I have had experience with second swapping areas under uport and isc.
I don't presently have one active.  With uport the first
area (on drive 0) is hard-coded into the kernel, and you must
leave it in place lest unix die: you can't delete it after adding
the space on the second drive.

I encountered a peculiar behavior:  after hours or days, the listing 
produced by 
	ps -ef
would contain corrupted lines.  I was able to provoke this by
loading a large program, which presumably triggered the swapping
algorithm.  Presumably the kernel either clobbers
its process table while using the second swap area, or gets into
some sort of trouble with things that it stores there.  This spooky
experience led me to stick to drive 0 for microport swapping.

I haven't exercised isc as thoroughly, but I did partition some
swap space on drive 1.  By now, though, extra memory is sufficiently
cheaper and faster that I've left my swap area quite small (1 or 2 Meg)
with  8 Mb of ram.  This weekend, gnu emacs threw me an "out of memory"
error.  If it recurs maybe I'll activate that extra swap space.

-- 
Norman Kohn   		| ...ddsw1!nvk
Chicago, Il.		| days/ans svc: (312) 650-6840
			| eves: (312) 373-0564



More information about the Comp.unix.i386 mailing list