MPU Meeting Tuesday 9th April 2013
Security Enhancements
Once Stephen has made the adjustments which Alastair suggested to the
final report the project will be submitted for sign-off at the May development meeting.
Inventory
The familiarisation work continues.
Login Logs Viewer
CEG has agreed a Data Protection statement, so Stephen will now deploy and publicise the login logs viewer.
Sleep Enhancements
Beta testing continues.
Once the gconf component has been deployed we'll look into using it to disable the sleep feature of GNOME Power Manager, as people are enabling it and then getting confused between it and the sleep component.
Virtual DICE
Chris has been having trouble with
roaming.h
on his DIY DICE machine. It should work, but may not be quite appropriate. Stephen and Alastair had some suggestions to make as to what to use instead.
LCFG Client Refactoring
Stephen continues to
blog about his progress on this project.
rdxprof
has now been cleaned up and most of the code moved into the
LCFG::Client
module. He's now running the client code through
perltidy
and
perlcritic
. He aims to satisfy perlcritic to level 4, but also to run it at all levels of fussiness to highlight issues which it would be profitable to tackle.
Miscellaneous Development
DIY DICE now defaults to SL6.3.
Operational
- AMD/ATI Catalyst/fglrx
- We're still suffering fallout from the removal of this driver. In particular,
- We've had to add a cron job to force the reinstallation of the mesa-libGL packages to replace missing files which had been altered by the Catalyst driver package and were then deleted when that package was removed;
- In some cases the xorg.conf file was not automatically removed and needed to be manually deleted.
- xemacs
- Some people have been unable to start
xemacs
in recent days (tickets 61615 and 61576) while for others it's perfectly functional. We don't yet have a solution.
- New KVM Server
- A new KVM server for AT will be arriving soon.
- KVM Server Kernel Upgrade Issue
- Sometimes a kernel upgrade on a KVM server will cause no problems for its running KVM guests, but at other times when a KVM host has been rebooted to upgrade the kernel the guests which have been suspended for the reboot will not resume, giving error: Unable to read from monitor: Connection reset by peer. The solution is to delete the suspended session and do a cold boot. A session can be removed using the
managedsave-remove
virsh command. We have to conclude that KVM server upgrades may cause problems for guests so from now on we intend to do them only once guests have been shut down.
- 358 Kernel Possible I/O Issues
- There has been a report on the SL mailing list about possible I/O performance issues with the new 358 kernel. We're going to compare the performance of the new kernel with the current one.
- JCMB machines' primary interface
- We need to set the primary interface to
mh0
for our machines at KB, as recommended by the inf unit.
This Week
- Alastair
-
Add virtualbox additions by default at DICE level
-
Read source/trunk/BuzzSaw/docs
-
Ship gconf
- Educate individuals about inappropriate KVM guest sizes
- Inventory project playing
-
Start inventory project
- RT tickets
-
Wipe old machines (mousa, prague)
- Review outstanding LCFG bugs
- Look at testing latest kernel i/o performance on juice(?)
- Chris
- Look at shipping new xorg-x11-font-utils (and related packages) - RT 61615
- RT tickets
- Virtualised DICE image
- Review outstanding LCFG bugs
- Stephen
- Look at testing latest kernel i/o performance on juice(?)
-
Install kubelik as new student and general ssh server
- RT tickets
- Wipe old machines (
rydell, tobermory, trondra)
-
Review outstanding LCFG bugs
- Deploy and publish user log viewer (and blog article)
-
Second innovation day
- Carol
- carry out an audit to make sure that all guests are still required to free up resources for new guests, shouldn't leave unused, half-managed machines lying around
--
AlastairScobie - 09 Apr 2013