[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Should 470629 be RC?



On Sun, Sep 14, 2008 at 08:30:59PM +0100, Greg Kochanski wrote:
> I'd like to make the case that bug 470629 ought to be fixed in Lenny.

> The bug is that a process spends an unreasonable amount
> of CPU time searching, waiting for someone to plug in a chip card.
> Now, this may not seem important, but I did the math, and it turns out
> to be a surprisingly large waste of energy and money.

> On my system, a 2.8GHz P4, it uses 3% of my CPU, 24 hours a day.
> For me, this is responsible for a (conservatively estimated)
> 1-watt increase in power consumption.
> (On more modern CPUs, this ranges from 0.1W to perhaps 0.5W.
> I'll take a conservative assumption that most people are running
> modern efficient CPUs and that the impact on the average
> user is 0.3 watts.)

> Now, when you multiply that increase in power by the number of
> Debian users (counting distributions that use Debian), it gets
> big.   With 10 million users worldwide, we are looking at a total
> power consumption -- from this single program -- of several
> megawatts.   (Enough to power several thousand houses or a
> good fraction of a percent of a nuclear power plant.)

This is hyperbole.  Percentage of CPU time is not linearly related to % of
energy usage; and multiplying by the number of Debian users is *wrong*,
because most users will *not* have this package installed.  The
reverse-dependencies of libchipcard-tools are limited to med-practice and
libchipcard-libgwenhywfar47-plugins (which may be pulled in as a recommends
via aqbanking-tools or qbankmanager).

>From http://qa.debian.org/popcon.php?package=libchipcard, roughly 1000
systems, or about 1.3% of systems participating in popcon, report having
this package installed.

It looks like I had this package installed on one of my systems for some
reason - but the CPU usage according to top was on the order of .3%, not 3%,
for a much slower system, which, BTW, has no power management capabilities,
so the net power consumption from having this process running is
approximately 0.

But thanks for pointing it out, I don't need this daemon at all so I've now
removed the package from the affected system. :)

Anyway, it looks to me like a simpler solution would be to have the
libaqhbci13 package drop its Recommends: on
libchipcard-libgwenhywfar47-plugins to a Suggests:, which would
significantly reduce the number of users having this package installed,
causing their power consumption as a result of this package to instantly
drop to 0.

This should definitely not be treated as RC - but it might be allowable as
an "important" bugfix for this optional package and let into lenny on that
basis, if the fix is suitably clean.

-- 
Steve Langasek                   Give me a lever long enough and a Free OS
Debian Developer                   to set it on, and I can move the world.
Ubuntu Developer                                    http://www.debian.org/
slangasek@ubuntu.com                                     vorlon@debian.org


Reply to: