SUMMARY: Solaris 2.6 interrupt/performance question

Sean Walmsley (sean.p.walmsley@hydro.on.ca)
Wed, 10 Dec 1997 10:56:02 -0500 (EST)

Original question (summarized):

>As part of our 2.6 testing, we've noticed that 2.6 machines seem to be
>generating a lot more interrupts than under 2.5. This is noticeable
>using the performance meter as well as vmstat.
>Solaris 2.6
>interrupt total rate
>--------------------------------
>clock 112872362 100
>zsc0 1783529 1
>zsc1 2719104 2
>cgsixc0 22057 0
>lec0 9974521 8
>fdc0 45140 0
>--------------------------------
>Total 127416713 112

The general concensus was that the 2.6 kernel is keeping track of more devices
and thus showing a higher interrupt rate. None of the responses indicated that
the interrupt rate seen was abnormally high or should be a concern in terms
of performance.

Thanks to:

Glenn Satchell <Glenn.Satchell@uniq.com.au>
Casper Dik <casper@holland.Sun.COM>
Jim Harmon <jim@telecnnct.com>

for their prompt responses.

--
Sean Walmsley
Reactor Safety and Operational Analysis Dept,
Ontario Hydro (H11-G26)
Tel. (416) 592 4608  Fax (416) 592 4966
e-mail: sean.p.walmsley@hydro.on.ca