[4947] in Release_7.7_team
Re: 2004 Athena Cluster Usage draft is complete.
daemon@ATHENA.MIT.EDU (andrew m. boardman)
Thu Feb 24 17:42:43 2005
Date: Thu, 24 Feb 2005 17:42:20 -0500
Message-Id: <200502242242.j1OMgJRJ014436@pothole.mit.edu>
From: "andrew m. boardman" <amb@MIT.EDU>
To: jweiss@MIT.EDU
CC: wdc@MIT.EDU, release-team@MIT.EDU, hallisey@MIT.EDU, pbh@MIT.EDU
In-reply-to: <200502142308.j1EN8Gxu017646@the-other-woman.mit.edu> (message
from Jonathon Weiss on Mon, 14 Feb 2005 18:08:16 -0500)
>Is there a reason you used spring terms, rather than fall terms, which
>would have allowed you to used data from the term that just ended?
I don't remember details, but according to the 2002 report, there was a
bug in the release in autumn 2002 that caused the statistics to be
incorrect. That said, I've just run off a series of graphs using the
data we have, and it looks in line with expectations.
>It's interesting to note that w20-575 percentage usage is down despite
>the loss of 15 (greater than 10%) of its traditional seats. Or am I
>confused about when that happened?
Looking at this, which didn't exist before:
/afs/dev/user/amb/athstats/graphs/full.aot.autumn/W20-575.aot.ps
...the autumn 2004 usage on Mondays is lower than that for previous
years, but for all other days, it's higher, sometimes notably so.
Cluster usage overall is also up, which implies that we may have chased
people elsewhere. (I find that, even when machines are theoretically
open, they often have people/stuff/dinner sprawled over them and/or a
reallocated chair, keyboard, or monitor, or they're just busted.)
>the 37-318 multi-year average graph does not appear. (there's even a
>note about it, now that I look carefully). I fail to see why 37-332
>and 37-318 should be considered seperately, rather than as the single
>cluster they really are.
The current definition of what's a cluster follows the definitions in the
source data. As for why there's no multiyear graph for 37-318, gnuplot
barfed on the lack of data from Spring 2001; now fixed. (Poke around in
/afs/dev/user/amb/athstats/graphs for more that you probably ever wanted
to see.)
