Re: strange load peaks with recent kernels

rnzll3!sys3.pe1chl!rob@relay.nl.net
Sun, 28 Apr 1996 09:02:06 +0200 (MET DST)

According to Greg Patterson:
> It is simply a change in the way load averages are calculated.. The
> change was made some time ago.. Probably the 70 series of kernels. It
> doesn't affect performance.
>
> It occurs for everyone... Load average calculations are more sensitive
> now (or maybe more accurate?).
>
> Anyways, right now I have a .34 average and thats about normal these days.

I think it is a bug...
It clearly is nondeterministic. Sometimes the load average sticks at
.5, sometimes at 1, for you it apparently sticks at .34
It can suddenly ramp up and ramp down, sometimes to 0 and sometimes to
integer values. From the above values it kind of looks like it sticks
at "n" or "1/n" where n is an integer.

I think this has to be fixed, the load average is not just "some toy"
but it is (or at least was) a measure for system load that can be used
by some processes to make decisions about starting background jobs, etc.

Rob

-- 
+------------------------------------+--------------------------------------+
| Rob Janssen         rob@knoware.nl | BBS: +31-302870036 (2300-0730 local) |
| AMPRnet:       rob@pe1chl.ampr.org | AX.25 BBS: PE1CHL@PI8WNO.#UTR.NLD.EU |
+------------------------------------+--------------------------------------+