Steven M. Bellovin wrote:
> 
> My experience was *way* before timecounters...  It was a Dell Precision
> 410, originally running 1.5.x.  I *think* the problem happened some time
> after boot (i.e., I never saw it in single-user mode at boot), but didn't
> stop when I returned to single-user.  But it's a few years since I looked
> into it, so I could easily be wrong about that part.
> 
> 		--Steven M. Bellovin, http://www.cs.columbia.edu/~smb
> 

Oh :-)

Well, hrm. I did just find this:

---
(root@bone)/usr/src/sys/arch/i386/conf# vmstat -t -c 5 -w 1
procs            memory
ru dw pw sl sw   total-v  active-v  active-r vm-sh avm-sh rm-sh arm-sh free
  2  0  0 19  0     55024     26409     25125     0      0     0      0 
  8914
  2  0  0 19  0     56272     27657     26373     0      0     0      0 
  7666
  2  0  0 19  0     55328     26714     25430     0      0     0      0 
  8610
  2  0  0 20  0     56325     27718     26434     0      0     0      0 
  7613
  2  0  0 20  0     57129     28522     27238     0      0     0      0 
  6809
(root@bone)/usr/src/sys/arch/i386/conf# uptime
  4:56PM  up 18 days, 18:02, 2 users, load averages: 4.27, 4.07, 3.15
---

Surely the load average here should be ~2 (as shown by "ru", which I 
understand is the current number of processes on the run queue) and not 
~4? I was just building a kernel at the time.
-- 
Mark Cullen <mark.r.cullen@gmail.com>