These were accepted.
Craig has asked around all the likely suspects, but hasn't been able to find anyone to own up to being consulted about this. He'll chase this up with Dave Berry directly.
Graham hopes to do this next week.
This seems to be worse on machines with slow RAID controllers and on desktops. There's a suspicion that it might be a kernel bug. There's another kernel upgrade pending on <develop> machines which might help throw more light on the matter.
MP-unit have made a start on this. The intention is that it should be undefined by default for desktops, or set to low/medium/high for servers. They're then expecting to have a script which can shut down classes of servers at a site (with appropriate safeguards!).
There doesn't appear to be an easy way to do this. It could be done for individual logged-in users' home volumes, one by one.
Defer to September as no BC before then.
The list has gone to CCL (North), and the uplift is expected to be on 4th July. Please don't add anything to B.03 until after then.
After the uplift, we all then need to review the remaining stuff in B.03 to see whether any of it still really does need to be kept.
(This action done; new action created.)
What do we do about old discs with odd connectors? Do we just trust that they will be shredded? We do have the option to send someone, by arrangement through Waste Management, to observe the uplift and subsequent destruction.
ICSA are happy for this to go ahead, provided they get a bit of notice beforehand.
Blog articles suggested:
Report from Computing Executive Group
Alastair hadn't been able to find proceedings for previous conferences. He asked for notes of interest in attending, but proposed waiting to see the list of papers before committing.
Reports from units
There should never be more than one "wire" header in any profile, so the best way would be to use one common macro both to protect the headers and to #warn in case of multiple inclusion. (Looking at RT#62330 and live/mpu-kvm-server.h later suggests that it might be useful for that protection macro to indicate which wire header it was set in.)
Stephen will be adding the DICE_STICK_WITH_SL63 macro automatically to profiles, as usual with this kind of upgrade.
This version won't be including kx509 by default.
It would probably be a good idea to change the root password on this (and similar) machines.
Topics for discussion
Recording roles/groups for machines:
Graham noted that Sun Java 1.6 would be going in SL6.4; 1.7 will remain (and become default).
The next meeting should be on Wednesday 10th July at 10am in IF-4.31. Given that it is holiday season and a number of people will be away, we'll decide nearer the time whether to hold it or not.
Please contact us with any comments or corrections.
Unless explicitly stated otherwise, all material is copyright The University of Edinburgh