vmstat -i weirdness

[Erik Trulsson]

On Tue, Jan 02, 2007 at 02:08:44PM +0400, Tofik Suleymanov wrote:
> Hello list,
> 
> looks like `vmstat -i` acts weird on my machine after being 12-15 hours 
> uptime.Here is the iutput of `vmstat -i`:
> 
> > vmstat -i
> interrupt                          total       rate
> irq1: atkbd0                        6813          0
> irq9: acpi0                         5397          0
> irq12: psm0                        73782          1
> irq14: ata0                        74209          1
> irq15: ata1                           47          0
> irq18: uhci2                           1          0
> irq19: uhci3 ehci0                     1          0
> irq21: iwi0                        35139          0
> cpu0: timer                    105315537       1999
> Total                          105510926       2003
> >                                                   
> 
> Strange is that for example atkbd0 has rate of 0, but total interrupts 
> count of atkbd0 is growing.
> Machine runs FreeBSD 6.1 RELEASE p11 with  pretty common  kernel.
> 
> Is this known behaviour ?

That is known and expected behaviour.  It is just a round-off error due to
the use of integer division.
'rate' is the average number of interrupts/second calculated over the
whole uptime of the machine.
Since you probably press a key on the keyboard less than once per second (on
average) this means that rate < 1 for atkbd0 and gets displayed as 0.

If floating point values were used to display the rate you should see a
value of maybe 0.13 for atkbd0.


-- 
<Insert your favourite quote here.>
Erik Trulsson
ertr1013 at student.uu.se