Greetings;
In attempting to profile last nights dump so I could get a better feel for how
amanda was doing, specifically wanting to find out how much data per minute
was being moved, I found that while it did draw a plot, some of it was pretty
nonsensical. Hundreds or errors were output to the shell I ran it from, and
several of the traces appeared to be basing their zero point on the bottom of
the plot window rather than the assigned area of the screen as noted in the
scale marks at the left, and the area assigns on the right.
I assume this is because the logfile format now used in amdump.x logs has
changed without a corresponding update to amplot.
Can this be looked at?
I am at the moment, playing canary with the cpu scheduler in the linux kernel,
trying patches from Ingo Molnar, and it would be very helpfull to have plots
of such as the write speeds achieved to the vtape available, as some
scheduler options can either smooth the profile I see in the gkrellm display,
or make it look a lot like the teeth of an upturned comb. This may also be a
finer grained graphic timewise than can be obtained with amplot, but
gkrellm's time resolution there is maybe 1 second at best and it doesn't
attempt to smooth to an average either. I'd love to be able to see a write
speed plotted over time where one could actually pick out the individual
files in the time plot.
All of this data can help smooth the desktop experience of running linux, and
what we've found so far has managed to get rid of the huge majority of the X
and kmail composer lags, and in one case where I use some really big hubble
pix for screen backgrounds that used to take about 10 seconds to render when
that screen was switched to the first time after a reboot, is now about 3.5
seconds, so we are making progress in this 'usability when busy' dogfight.
--
Cheers, Gene
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Sin has many tools, but a lie is the handle which fits them all.
|