[6038] in Release_7.7_team
Re: Fwd: Putting display to sleep on Athena
daemon@ATHENA.MIT.EDU (Jonathon Weiss)
Mon Jul 28 18:16:03 2008
Message-Id: <200807282215.m6SMFN9x025891@speaker-for-the-dead.mit.edu>
From: Jonathon Weiss <jweiss@MIT.EDU>
To: Greg Hudson <ghudson@MIT.EDU>
cc: William Cattey <wdc@MIT.EDU>, release-team@MIT.EDU
In-reply-to: Your message of "Mon, 28 Jul 2008 16:39:33 EDT."
<1217277573.12433.35.camel@error-messages.mit.edu>
Date: Mon, 28 Jul 2008 18:15:23 -0400
X-Spam-Flag: NO
X-Spam-Score: 0.00
> On Thu, 2008-07-24 at 13:13 -0400, William Cattey wrote:
> > Q1: Leaving aside these concerns as policy issues, what is the
> > technical solution to switching systems over to putting the display
> > to sleep after 15 minutes?
> >
> > Q2: If the policy keeps cluster systems from sleeping, what is the
> > technical solution to setting just Private system to do this?
> >
> > Q3: Is there a particular approach that we should advocate?
>
> Just a note that I have read this, but do not know the answer to these
> questions offhand and do not have the time to put into researching them.
1)
Isn't "monitor goes into powersave mode" something we're going out of
our way to disable in Athena 9? Specifically,
/etc/athena/login/Xreset contains:
/usr/X11R6/bin/xset -display :0.0 -dpms
I believe we would get power saving after the default amount of time
(which may or may not be 15m (or possibly 24-27-30m (standby-suspend-off).
Alternately, replacing it with a line like:
/usr/X11R6/bin/xset -display :0.0 +dpms 900 1020 1200
would suspend after 15 minutes and be of by 20
Note that the defaults are form the solaris man page. The linux one
doesn't seem to indicate what the deafults are.
This probably also means we'll get this for free in Athena 10 by
virtue of having dropped this code.
2)
I believe a private workstation owner could edit
/etc/athena/login/Xreset though they would risk having their change
overwritten whenever we replaced the athena-ws package.
3)
As for the ability to see what's not displaying a logo, there are some
failure modes where the logo is up and displayed, but the machine is
hung hard and needs power-cycling. That's not to say that the case of
stuck in the middle of the boot process (generally becasue someone
unplugged the net cable) isn't more common. However, just displaying
the xlogin idle screen isn't a guarantee that the machine is working.
The work-around is that they have to walk the row of machines hitting
the shift key and then look at them, but tey'll end up noticing more
problems as a result, so it may well be somethin they should be doing
anyway. How much of a burden this is will depend on how often they
walk the clusters looking for down machines (as opposed to being
notified of them by users or automated reports). It's probably worth
re-visiting the issue with John Guy and company.
Certainly, the windows machines in the clusters sleep their displays
after a period of idle time. I don't think users will find the
concept foreign. It's possible that we'll need to do some education
since they aren't used to *Athena* machines sleeping the display, but
I don't think this is a huge deal. Actually, their first reaction
might be to pound on the keyboard (I think mine would be) to see what
was wrong with the machine, which would wake up the monitor and solve
the problem. A few temporary posters might help the issue too, but
I'm not positive they would be needed.
Given MIT's "walk the talk" push around energy, we really are way
behind on this particular issue. Machines have been able to sleep
their monitors for how long? It must be close to a decade. I think
we should bite the bullett and make the change. One option is to wait
until Athena 10 comes out but there is some risk of someone calling us
on completely failing to walk the talk. Given that it is currently
the summer it's a pretty good time to make the change too.
Jonathon
