[6174] in Release_7.7_team
Athena Release Team 1/21/09 Notes
daemon@ATHENA.MIT.EDU (Bill Cattey)
Wed Jan 21 17:35:17 2009
From: Bill Cattey <wdc@MIT.EDU>
To: release-team@mit.edu
X-Spam-Flag: NO
X-Spam-Flag: NO
X-Spam-Score: 0.00
X-Spam-Score: -5
Content-Type: text/plain
Date: Wed, 21 Jan 2009 17:35:42 -0500
Message-Id: <1232577342.7437.121.camel@lakota.mit.edu>
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit
Present jdreed, broder, jweiss, rbasch, wdc, amb, alexp
Actions:
1. Questions for team to review:
* Will we re-number low gids and low uids?
If so, what transition work is needed?
There are 51 UIDs below 200, and that seems like a lot of people to
move. There are 24 groups below 200.
There actually are low UIDs that have problems in Athena 9, but we've
been ignoring the problem.
UIDs are hard because AFS ACLs are keyed to UIDs.
Group 101 for group "mit" is the big problem.
Low uids and gids will not work for Athena 10 because they conflict with
current standard practice for creating local accounts and system groups.
Crawling AFS to perform the re-numbering would require writing code.
(Would probably be based on the existing janitor code, and it takes
about a week to actually crawl AFS.)
The work would be:
Confirm janitor can be co-opted to this task (jweiss talks to zacheiss)
Identify range of uids assigned by moira (jweiss)
Produce documentation to help those few people needing to chown or chgrp
files on local disk after their UID or GID changed.
Identify the affected persons.
Have the janitor crawl AFS to identify the affected ACLs.
The transition would be:
Send email warning the affected person of the change.
Change their UID (or GID) and update their affected ACLs.
Next step: Investigate the janitor and the affected users, ACLs.
* Shall we implement the proposed merge of Debathena and Athena10?
The proposal is:
IS&T takes over the hostname debathena.mit.edu, and takes over
administration of the debathena.mit.edu web site and apt repository.
The packages under Athena 10 are the IS&T Supported products.
The other packages, though living in IS&T administered
debathena.mit.edu, are un-supported by IS&T.
All SIPB debathena folk, as collaborators will retain write access to
the debathena.mit.edu web site.
SIPB debathena folk (non-employees) would get write access to the apt
repository upon agreeing to a to-be-defined set of rules and guidelines.
The merge will not take place until SIPB signs off on that set of rules
and guidelines.
(This merge has the advantage that IS&T gets rid of the problematic
build script to pull down source, build it, and deploy it via the
separate Athena 10 repository.)
* Does Athena still want to enforce blackout dates between drop date and
the end of term for patches offered by the Athena10/debathena
repository?
Tabled to next meeting.
* Future of Open Office Locker: Should we now follow what Ubuntu
releases? When should we go live with OOo 3.0? Should it be the
default?
alexp says that OpenOffice 3 should be made available sooner rather than
later. It's available on Athena 9.4, and he expects there to be demand
for it.
As a general rule, we prefer to stay in sync with what is the default
for Ubuntu, however OpenOffice 3 is not there. It is expected to arrive
in the backports repository on or about 27 January.
Decisions: We will make the OpenOffice 3 backport package available.
Athena 9 and Athena 10 will both switch to OpenOffice 3 as the default
version at the beginning of the spring term (early February 2009.)
* Should bash be the default shell for incoming students? If so, what
transition work is needed?
This is very easy to do from a technical standpoint.
Decision: Yes, we will flip the switch approximately now to affect
incoming pre-frosh.
jdreed owns checking in with Oliver and the other affected support stake
holders, updating documentation and then telling Garry to flip the
switch.
* Is the Athena 10 LiveCD an appropriate substitute for an Athena
10 VM image?
Yes, for now.
2. Re-visiting Athena 10 Kiosk mode:
What is the scope of need?
The requirement is for Register, but we think that offering the whole
web might be desirable.
What do we see as the exposure?
We hadn't thought it through very carefully, and were concerned someone
would have a problem with it. But looking at the history of other
publicly available desktop systems we think this won't be a problem.
So we will enable full web access for Early Cluster release to learn if
it poses a problem. If it poses a problem we can shrink the access down
to the specific sites such as Register.
3. Punch List to Athena 10 go-live:
Comprehensive installer: The difficult work of the LVM change is done.
A bit more polish is needed before the installer is complete.
Create debathena-thirdparty package: To be done after installer
complete.
ATN-1 Improved session names:
ATN-4 Greeter cosmetics: DECISION: Release team likes jdreed's change
(screenshot-alt.png) that does have the maroon background athering
around the white portions of the login box and owl logo.but Will go
with
ATN-9 skip_: Awaiting the work, but not really a blocker to Early.
ATN-10 attach/detach: amb did take FUSE changes. Awaiting broder update
of command utilities. Can be slipstreamed after Early go-live. Not a
blocker.
ATN-11 (sudo in clusters): status of broder fix to sudoers. Blocking
resolution of problematic GID 101. This issue deemed not a blocker to
Early by consensus.
ATN-12 Clean up local accounts: awaiting test from someone other than
jdreed.
ATN-13 Fix X configuration: we have a work around so not a blocker.
Remaining blockers: Installer, ATN-1, ATN-4.
wdc owns bringing ATN-1 and ATN-4 to closure.
jdreed will identify 20 candidate machines in W20, 56, 12, 66, and cook
up a sign for them.
wdc owns working with bmurphy and the Hotline team to do a roll-out.
Target Athena 10 dates:
Comprehensive installer ready for use by ridgway and wdc by Tue Jan 6.
Go-live in Early Cluster two weeks (plus holiday) later Wed Jan 21.
4. Check in on IAP Patch Release and follow-through.
We deployed the power saving code. No complaints so far.
We deployed Firefox 3, and it required a follow-up fix for printing,
which was also deployed. Solaris broke emacs-style editing in Firefox.
We have a work-around, and plan to leave it broken unless we hear
further complaints.
There was an overload hiccup because of a bug in the update desync code.
If we discover need to make another patch release, we'll incorporate a
fix for this bug that patch, but won't craft a patch solely for this
bug.
5. Prep for TAP
wdc and others are doing the prep.
