[PLUG] vmstat question

Matt Alexander lowbassman at gmail.com
Sun Oct 17 08:09:02 UTC 2004


On Sat, 16 Oct 2004 22:42:34 -0700 (PDT), Elliott Mitchell <ehem at m5p.com> wrote:
> >From: Matt Alexander <lowbassman at gmail.com>
> > Can someone explain how to track down which processes are waiting to
> > run when listed by vmstat?  For example, vmstat might show 15 procs
> > waiting for run time.  How can I find out what those processes
> > actually are?  Or more importantly, how can I determine what is
> > bogging the system down?
> > Here's a sample output from vmstat:
> >
> >    procs                      memory      swap          io     system      cpu
> >  r  b  w   swpd   free   buff  cache   si   so    bi    bo   in    cs us sy id
> > 17  0  1 310860   6924   1264  10564    0    0     0     0  123  1606  1  1 98
> 
> Wrong tool for this, run `top`. Default sorting order is by usage of
> processor time. Processes trying to use lots of processor time are
> obviously candidates you're looking for. Depending on the flavor of top
> pressing 'M' will tell it to sort by resident memory usage. By forcing
> the system to swap, processes using lots of memory can also slow things
> down.

top was the first program I ran before vmstat and it didn't show
anything unusual.  The CPU is close to 100% idle all the time.  I'm
guessing it's memory bound and the swapping is slowing things down. 
vmstat shows the si and so values to be almost identical to the bi and
bo values, so this leads me to believe that nearly all the disk
activity is swap related.  Does this sound correct?
Thanks,
~M




More information about the PLUG mailing list