[2430] in Release_7.7_team
Minutes of 2000-09-27 release-team meeting
daemon@ATHENA.MIT.EDU (Greg Hudson)
Wed Sep 27 13:58:29 2000
Date: Wed, 27 Sep 2000 13:58:19 -0400 (EDT)
Message-Id: <200009271758.NAA25019@small-gods.mit.edu>
From: Greg Hudson <ghudson@MIT.EDU>
To: release-team@mit.edu
Attending: ajfox ghudson othomas rbasch miki amb jweiss wdc
1. 8.4.13 patch release
Since this is a big patch release, we will wait a week and release it
on the evening of October 10.
2. xlock
As of 8.4.13, /usr/athena/bin/xlock now activates or runs xss. It
would be nice if we could start telling users to run xlock instead of
the more obscure "xss-command -lock". But, since Linux doesn't ship
an xlock, we don't have a /usr/athena/bin/xlock there.
The consensus was that we should make a /usr/athena/bin/xlock be
present on Linux. Greg will take care of this.
3. SIPB installer
The SIPB installer currently sets PUBLIC=false by default. SIPB
thinks this is the right thing but is willing to do what the
release-team wants. release-team has no strong opinion on the topic.
We've also noticed that the SIPB installer doesn't do everything the
Athena phase2 script does, like turn off the console logins on the
VCs. Since leaving the VC logins enabled is not a variation
documented in the PWOG (and thus not a supported customizatino),
that's a problem, and either the SIPB installer or the PWOG should be
fixed.
4. Linux issues
Greg noticed that dm doesn't always deal properly when X fails to
configure. Instead of falling back to a console login, it dies on an
X error and keeps respawning until init decides it has been restarting
too often.
Lou has noticed some machines where X fails to come up (but which
don't exhibit the above failure mode) and have had to be reinstalled.
We don't have enough information to fix this problem; jweiss and amb
have notified Lou that they are willing to look at machines in this
state.
Lou also noted that verification is slow. Possible ways of addressing
this issue:
* We might be able to make disks on cluster machines run
faster using hdparm.
* We could install less stuff. (Probably not an option for
8.4.)
* We could print a message saying that this will take a while
(possibly with the starting time).
* We could display a progress meter. We probably can't make
rpm -Va do this, but we could maybe do it if we ran rpm -V
on each package individually. That might slow down
verification even more, but maybe not perceptibly.
Greg will look into the right answer.
There is also the problem of machines coming up failing to fsck, which
we haven't solved.
5. Red Hat 7
Red Hat 7.0 is out. Some changes which we might have to think about:
* It uses XFree 4.
* It replaces inted with xinetd, which uses an
/etc/xinetd.conf instead of an /etc/inetd.conf.
* It uses a new version of rpm with a new preferred db format,
although it supports both db formats.
amb should schedule some time before February to work on the Red Hat 7
upgrade.
