Re: /proc/loadavg disagrees with top and ps
Reid Priedhorsky <reid@reidster.net> said on Thu, 05 Aug 2004 20:13:42 -0500:
> Hmm. So, the general consensus is that it's not a problem; and it
> certainly doesn't seem to affect interactivity or performance at all. It's
> my home box, not a server or anything, and it normally has very low loads,
> 10-15% maybe when I'm using it and essentially zero when I'm not. There
> shouldn't be lots of processes doing I/O.
>
> The high load dropped back to normal shortly after I posted.
>
> I'm still interested in tools that would tell me what processes are doing
> I/O, or whatever. It's unnerving for things to being going on with my box
> that I don't understand.
`ps axf` lists in the 3rd column the state a process is in. If you
have a process in 'D' state, it will contribute a value of 1 to the
load (as will 'R' - but then it is taking CPU and will appear in top's
output).
If it stays in D for long continuously (as opposed to intermitently
and for a few seconds - eg. while accessing the disk), then there is
probably a kernel bug involved somewhere.
If however, the load goes away after some time, maybe it is not
something to worry about. Were you waiting for slow IO from a disk or
floopy, or maybe listening to music on a bad CD? Any oopsen in your
syslog?
--
TimC -- http://astronomy.swin.edu.au/staff/tconnors/
If I sit here and stare at nothing long enough, people might think
I'm an engineer working on something.
-- S.R. McElroy
Reply to: