KDE is preventing laptop disk spindown by incessant configfile-writing
Date: Wed, 19 May 2004 11:40:24 +0200
From: Sylvain Joyeux <email@example.com>
I have the same problem here (writing of kdeglobals and korgacrc for no
obvious reasons). I don't have time to check the KDE sources.
See Riku Voipio's recent message (and my response) in this thread.
His suggestion fixed the problem. (The solution was to find
korganizer in the taskbar, disable alarms, and make it quit.)
Tell me: you do want to _use_ your laptop, don't you ? If not, just put it
into sleep :p. If you do, then you'll save data, read, whatever and -
eventually - your HD will spin up every 5 seconds because of the ext3 commit.
No, it won't. See my response to Riku. Getting rid of korganizer's
write-once-a-minute behavior let the disk spin down and -stay- down.
And I didn't have to do anything to ext3's commit interval (e.g.,
touching /tmp/foo does spin the disk up instantly, so writes to the
filesystem are still happening immediately).
I do agree that KDE shouldn't write its config files, but I don't see the
point of (desesperately) trying to fix something which is obviously not the
way you'll use your laptop (or at least, it is not the way I use mine ;))
If someone asks for help solving a problem, it's generally not very
helpful to anyone (the asker, the answerer, or the peanut gallery)
to insist that they're not solving a real problem just because it's
not -your- problem.
I often use my laptop to take notes in a meeting, or to read mail out
of a single large RMAIL buffer, etc. In those cases, there's no point
to having the disk spun up (and one of these laptops also has the
loudest disk I've ever heard, so it's annoying as well). Since I got
this behavior with no effort under Mandrake 8.2, I figured it should be
attainable under Debina. In fact, having now nuked korganizer, it was.
If you do really want to fix that, you should take a look at tmpfs. I think
the best way to achieve what you want is building a .kde/share/config in
tmpfs, saving it on disk on shutdown, reading it from disk on start.
An interesting idea. Don't need it this time, but maybe some other
time. (I was speculating on doing something like this when I first
noticed the problem, but was hoping it wouldn't be necessary...)