[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: mkswap: warning: truncating swap area to 2097144kB



[Please trim replies]

On Tue, Aug 20, 2002 at 12:37:49PM +0200, Yildiz, Murat wrote:
> then the rule (don't even know if it's a rule or thesis) Swap area=Physical
> Memory x 2  is dead , or?

The 2x rule was only ever an approximation (aside from early 2.4
kernels[1]).  I think the real question is 'Will I ever have more than
1GB of stuff loaded at once?'  I know I never would, on a desktop at
least.  This one's been up for 20 days, emacs, galeon, gnome, sawfish,
piles of Eterms, plus the usual miscellaneous stuff, and it's only using
~90MB of RAM (out of 128MB) and ~150MB (out of 768MB (related to [1]))
of swap.

Incidentally, what is the deal with the 'SIZE' column in top?  It seems
to undergo unbounded growth for some processes, especially for Galeon
and X.  It seems to fairly unrelated to the %MEM column.  I had an X
crash yesterday, and things certainly felt a lot more responsive when I
started X again; it's SIZE was back down in the teens of MB.

IIRC the SIZE of X was bloated by X loading images for running programs,
but shouldn't this be garbage collected somehow?

[1]  IIRC, early 2.4 kernel VM systems did not reclaim swap space when
something was swapped back in, so you eventually had a copy of your
entire RAM stored in swap.  Thus, you needed at least 1xRAM swap for it
to have any effect at all, 2x for something useful.

-rob

Attachment: pgpDRe0G4WLAC.pgp
Description: PGP signature


Reply to: