Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 18 May 1996 11:59:10 -0600 (MDT)
From:      Danny Dulai <nirva@mail.zynet.com>
To:        freebsd-questions@freebsd.org
Subject:   VSZ, RSS, and -STABLE
Message-ID:  <199605181759.LAA03434@ns1.zynet.com>

next in thread | raw e-mail | index | archive | help
what exactly does the VSZ column in ps aux mean?  I am told
that its the total ram a process takes up and RSS is how much
memory is in the core, but my ps aux output shows about half
the processes with a larger RSS than VSZ.

here is a sample:
USER       PID %CPU %MEM   VSZ  RSS  TT  STAT STARTED       TIME COMMAND
nirva     5272  0.0  0.0   556  436  p6  TW    9:16AM    0:00.21 vim main.c
nirva     5471  0.0  1.5   344  448  p4  I+   11:08AM    0:00.04 man ps
nirva     5472  0.0  0.6   492  180  p4  I+   11:08AM    0:00.01 sh -c /usr/bin
nirva     5474  0.0  1.5   216  444  p4  I+   11:08AM    0:00.18 more -s

I have also been told that VSZ is the total swap space availible.
that doesnt make sense either since when i do a clean boot,
ps aux shows a lot of VSZ, but pstat -s shows no swap in use.

here is a sample:
Device      1K-blocks     Used    Avail Capacity  Type
/dev/sd0s1b     65536        0    65472     0%    Interleaved

(thats from the same time as the above ps aux)

I've also been told that VSZ is how much a process allocates
and RSS is how much its actully using. i assume this has to
do with demand paging so you dont actually use the ram until
really use it, but im not sure.

someone please explain this RSS/VSZ thing to me, im trying to find a 
ram problem that ive been having since i went from -release
to -stable, i run out of ram on machines that i didnt
with -RELEASE but now am with -STABLE.  processes seem to go
into deep sleep before the system dies, and all those sleeping
processes take up tons of ram.

I have 64 megs ram and 64 megs swap in this machine, 48 megs
ram and 64 megs swap in another, and 32 megs ram and 64 megs
swap in another, all went from -RELEASE to -STABLE, all
experiecned the problem.

my reason for the switch was that if i used setsockopt()
with the IP_HDRINCL option under certain circumstances, i got
a kernel panic with -RELEASE, this was fixed from panicing in
-STABLE, now it just returns an error when in use sendto()
after using IP_HDRINCL. *sigh*


------------------------------------------------------------------------------
Danny Dulai                                             Feet. Pumice. Lotion.
http://www.ishiboo.com/~nirva                               nirva@ishiboo.com
------------------------------------------------------------------------------



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199605181759.LAA03434>