I just noticed that on one machine running 2.0.0 the load average
is too high by one. This is after about 2 days of uptime:
$ cat /proc/loadavg
1.11 1.06 1.02 2/44 26547
The system is idle, CPU usages of all processes are below 1%
according to "ps" (and only "ps" is in the R state). It seems
that someone is not decrementing the nr_running variable (or
incrementing it twice).
I think it has been reported before (late 1.3.xx kernels).
I'm reporting it here to let people know that it's still there
in 2.0.0. Not a very serious problem, but...
Marek