call for testers: altq in current

Eygene Ryabinkin rea-fbsd at codelabs.ru
Wed May 2 11:54:31 UTC 2007


Nate, *, good day.

Sorry for the previous posting: I had messed the things and posted
the old message to the list.

Mon, Apr 30, 2007 at 03:27:26PM +0400, Eygene Ryabinkin wrote:
> > > On the new code but without loading cpufreq and leaving the freq at 2200
> > > Mhz, do you get the right numbers?  Are they constant?
> > 
> > Monday will reveal the things. Will post an update.
> 
> Was not able to test the things on Monday. But will try to do it
> on this week.

OK, with new code (patched with my patch) and cpufreq the download
rate is more-or-less stable. But it is 4x larger than the limit I'd
set in the pf.conf. And there are times when the rate goes straight
to the maximal link bandwidth overriding any altq statements. But
later it returns to the stable (4x larger) figure.

Without cpureq I have stable rate, but it is again larger than the
value specified in the pf.conf. At the time of testing I got the
multiplier 33 (for the 3Kb specified in pf.conf I got 100 Kbit as
measured by ifstat and fetch), but I feel that reboot or kernel
recompilation will give me some other multiplier. And I still have
the bumps of the bandwidth to the link-rate limit.

To conclude: the frequency callback is doing its work now (with my
patch). Since I have strange results with the real rates I will
continue my investigations and will try to post updates when I will
discover something.
-- 
Eygene


More information about the freebsd-current mailing list