[7205] in testers
afs kernel oopsing
daemon@ATHENA.MIT.EDU (Kevin Chen)
Sun Jun 26 20:22:28 2005
Message-ID: <42BF46B8.40005@mit.edu>
Date: Sun, 26 Jun 2005 20:22:16 -0400
From: Kevin Chen <kchen@MIT.EDU>
MIME-Version: 1.0
To: testers@mit.edu
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
After 7 days of uptime, scyther.mit.edu still hasn't kernel oopsed
again. Since it sounds like this wasn't observed in alpha or beta, and
dodecahedron.mit.edu also had the same problems as scyther.mit.edu, it
seems like it might somehow be related to the number of users,
processes, or resources being used.
Before 9.4 on scyther.mit.edu, about 7 people would usually be
simultaneously logged into scyther.mit.edu remotely, all with screen
sessions with owl (and often pine) running. Pretty much the entire 1 GB
of RAM would be used (according to top), and 400-700 MB of swap would be
used.
Since the last kernel oops on scyther.mit.edu, many people started using
other machines, and now about 2 people will be logged in remotely (plus
me locally), the 1 GB of RAM is used, and as of right now, only 1.3 MB
of swap is used.
I don't regularly use dodecahedron.mit.edu, so I'm not sure what kind of
usage it gets, but I believe it doesn't have that much RAM and usually
has plenty of users.
I'm not sure if these statistics are useful, but perhaps it can provide
some insight into what's causing this problem and why it hasn't been
seen before.
--
Kevin Chen
http://www.sneswhiz.com/
