Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 19 May 1999 14:03:17 -0500 (EST)
From:      John Fieber <jfieber@indiana.edu>
To:        Greg Quinlan <greg@qmpgmc.ac.uk>
Cc:        freebsd-stable@FreeBSD.ORG
Subject:   Re: Memory leaks & kernel panic/reboot & ahc reboot 
Message-ID:  <Pine.BSF.4.10.9905191337230.65957-100000@fallout.campusview.indiana.edu>
In-Reply-To: <00ea01bea21e$3bad8a20$380051c2@greg.qmpgmc.ac.uk>

next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, 19 May 1999, Greg Quinlan wrote:

> Previously Mike Smith wrote:
> >The Active/Inactive/Wired/Cache/Buf/Free statistics refer to the
> >disposition of physical memory.  They are unrelated to swap usage.....
> 
> Physical memory unrelated to swap.... that's a new one!

You are misinterpreting what top is is reporting.  Of course
physical memory and swap are related but the particular numbers
in question do not have anything to do with swap.  As an analogy,
an oil pressure gauge and a fuel gauge are related in that they
both indicate some operational parameter of an engine, but the
oil pressure gauge doesn't tell you anything about how much fuel
you have left, it tells you about the oil pressure.  Likewise,
the Active/Inactive/Wired/Cache/Buf/Free figures tell you
absolutely nothing about swap, they tell you about physical
memory.

Unfortunately the correct interpretation in this case isn't as
obvious as it is with the oil pressure gague.  As a general rule
on *any* system, you cannot interpret a reported number unless
you know how the number was calculated and you (recursively)
understand each element of the calculation.  These things are in
desparate need of careful documentation.  Incidentally, this is
also why you cannot directly compare figures between different
operating systems.  You sometimes can't even compare between
different versions of the same operating system.  The calculation
of "SIZE" and "RES" as reported by top changed between FreeBSD
2.x and 3.x for instance.

You claim to have evidence of a memory leak when in fact there is
absolutely no such evidence in the data you report. You have gobs
of free memory and your swap space has hardly been touched.  As
someone else mentioned, you would have to chew through about 84
megabytes before swap would even be touched. Then you have yet
another 120 or so before you run into serious trouble.

Whatever problem your system is having, a memory leak and/or
shortage is definately not it.

Good luck tracking down the real probelm.

-john





To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-stable" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.10.9905191337230.65957-100000>