[1160] in Release_7.7_team
Meeting notes, 12/2/97
daemon@ATHENA.MIT.EDU (Greg Hudson)
Tue Dec 2 13:53:07 1997
Date: Tue, 2 Dec 1997 13:53:01 -0500
From: Greg Hudson <ghudson@MIT.EDU>
To: release-team@MIT.EDU
We didn't talk about Athena-maintained lockers because Dan wasn't
ready.
bugs@mit.edu spam:
- There isn't much we can do to stop getting spam.
- If and when we start using a bug-tracking system, we should
make sure it supports some kind of filtering so that it can
try not to open bugs for spam mail.
- danw will investigate a spam-filtering service for menelaus.
(Which might not be the best place for a spam-filtering
service.)
time synchronization:
- We used to run timed on our machines.
- We don't want to use timed now, since it requires a server
for every subnet.
- Right now we rely on AFS (which doesn't work all that well)
along with a 12-hour fallback to gettime.
- NTP is the commonly used protocol on the net right now, and
is more precise and reliable than AFS.
- The new MITnet routers support sending NTP broadcast packets
to set the time, which is a nice network-efficient model.
- But we still need to be concerned about machines maintaining
a reasonable time if they're on a network which doesn't get
NTP packets.
- We don't think NTP will fight with AFS because of AFS's
two-second granularity. NTP will sort of fight with
gettime, but we can introduce a granularity argument to
gettime to prevent that.
- We probably still want a gettime fallback, since AFS still
has known problems with time-setting and not everyone's
network will have NTP.
- We should note that we are trusting the network group to
properly maintain time on the routers, and that that time
has to agree with time.mit.edu and with the AFS servers to
within a couple of seconds.
- brlewis will incorporate xntpd into the source tree.
- ghudson will take care of the gettime options.
RFP platforms:
- We don't want any new ones this year. Maybe for next year's
RFP, when we hopefully go back to having only one SGI
platform.
Release documentation:
- ghudson had questions about why Kevin was departing from the
system release notes/user release notes model. Kevin wasn't
here, but people thought that this was just to let people
know about the major new items in the release well in
advance of the testing cycle.
