[5845] in Release_7.7_team
Collecting Athena release features
daemon@ATHENA.MIT.EDU (Greg Hudson)
Tue Sep 18 15:05:52 2007
Date: Tue, 18 Sep 2007 15:05:09 -0400
Message-Id: <200709181905.l8IJ59TJ020593@equal-rites.mit.edu>
From: Greg Hudson <ghudson@MIT.EDU>
To: release-team@MIT.EDU
X-Spam-Flag: NO
X-Spam-Score: 0.00
As currently planned, Athena 10 will require a great many changes
like:
Problem: Athena machines keep their time synchronized with
time.mit.edu.
Current solution: We provide a utility to reset the time
(athena-gettime) and a build of NTP 4.1.1 (athena-ntp). The
athena-ntp package resets the time at boot or network startup, and
also starts up the NTP daemon to sync the time passively from
network broadcasts, which are present on most MIT subnets. The
athena-ws package sets up a cron job to reset the time twice a day
using gettime if it has drifted. AFS will also attempt to keep the
time synchronized to within two seconds.
New solution: We will ensure that the ntp and ntpdate Ubuntu
packages are installed. A athena-ntp-config package will configure
ntp to sync the time actively against time.mit.edu, which should be
a one-line change. AFS will continue to resync the time if NTP
somehow fails. We will continue providing the athena-gettime
package for compatibility but will not use it in any of our own
scripts.
That's a mildly complicated example, but it shows the kind of
reworking we'll want to do in order to get out of the business of
building all that stuff in "third". I believe the best way to
organize Athena 10 work will be for me to create some basic
infrastructure and then parcel out these problem-old-new tasks to
Andrew and Bob. (This should feel a bit like agile programming.) To
do that, I need to accumulate a list of problem statements. I'll be
making my own list, but it will be easy to miss some with the Athena
release being as old as it is. So, please brainstorm and send me
release features you think might not be obvious. Don't worry about
the current and new solutions; I'll fill those in.
We can probably ignore fiddly things like "Athena cluster machines
automatically work around a deficiency in RHEL 4's hardware support
for the XYZ device" and assume that we will have an distinct set of
those problems with Ubuntu 8.04.
Also, while "Athena can run software out of AFS lockers" is going to
be a problem statement, problems we solve with locker software (like
"Athena users can run Matlab") are out of scope.