MPU Meeting Monday 1st June 2009

Power Management Project

Stephen asked Chris to document how to debug sleep-related problems such as failure to resume from a suspend.

Alastair suggested upgrading the BIOS on the test machines to see if that made them suspend and resume any more reliably.

Sleep was tried with the i810 driver but the test machines didn't reproduce their earlier behaviour with i810, instead behaving considerably worse. Meanwhile Graham got permission to upgrade webots to a newer and less broken version, so (pending test confirmation) he expected that we'd be able to use the intel driver on 5.3 machines. He recommended switching the develop release to intel right away. Chris did this. Taking all this together, lcfg-sleep support for i810 now seems irrelevant and has been abandoned.

Alastair suggested that the component be rolled out to at least one student lab as a test.


Stephen has reviewed the code and passes it, although there are areas of doctrinal disagreement.

Architecture documentation is still required.

The two new servers have arrived. Alastair is waiting on AFS admin privileges. Stephen may be able to help him get these quickly.

AFS Component

Stephen has been trying to set up a test cluster. Various virtual machine environments were tried. VirtualBox in particular seems unreliable and difficult to use and wastes far more time than it saves. One database server is up and running. Another is needed, then at least one fileserver.


Nothing more from Craig so far so no progress. It was suggested that Chris draw up his own design/requirements document and pass it by Craig.

LCFG Server Refactoring

The project has been submitted to the June Development Meeting.

Miscellaneous Development

om disowning credentials
Fixing this problem will be easier after the code has had a general tidy-up, so Stephen has embarked on that.


The LCFG wiki has been upgraded
While he was at it Stephen added a handy bugzilla plugin which lets you quickly refer to an LCFG bug from the LCFG wiki. See for details.

Move dresden data to evo array
A partition has been made on the array - LUN 11 - and it's permitted to both dresden interfaces, but dresden can't see it. It appears to be an OS-level problem. Normally scli can be used to look into such problems but its RPM installation seems to be damaged on dresden. Alastair will look into repackaging scli.

SL 5.3 Upgrade
Alastair took this to CEG, which decided to push out the upgrade to SL 5.3 in 6 weeks (the stable release of 16 July), and to force reboots the week after that for any servers which hadn't already fully upgraded. Stephen has made an SL 5.2 package list for the lab machines which are being held back until the autumn.

Could we tidy up and close RT tickets please.

Server move
Our non fibre channel servers should be moved to a rack away from the fibrechannel equipment. Initially Carol will draw up a list of which servers have to move, then we'll exclude from that list any servers for which we want to retain emergency fibre channel capability.

Meeting room machines
Support put in a ticket (42239) recently asking for shutdown to be disabled on machines in meeting rooms. Alastair is happy with this idea as long as people can still reboot them. Stephen will look into it.

This Week

Alastair will:

  • Talk to Carol about the LCFG level work.
  • Book a virtualisation mini-talk slot with Alison.
  • Remind Craig about moving hawthorn to KB.
  • Work on rpmsubmit project.

Carol will:

  • Set up LCFG/inf level VM to monitor LCFG level.
  • List which servers need to move from the FC racks.

Chris will:

  • Help Carol with the testing release. DONE
  • Move dresden data to evo array.
  • Summarise the sleep project. DONE
  • Document how to debug sleep problems. DONE

Stephen will:

  • Tackle 'om' improvements.
  • Look into inhibiting shutdown on meeting room machines.
  • Continue AFS component testing.
  • Document the use of the AFS component.

-- ChrisCooke - 2 June 2009

Topic revision: r2 - 09 Jun 2009 - 08:34:00 - ChrisCooke
DICE.MPUnitMeeting20090601 moved from DICE.MPUnitMeeting20080601 on 02 Jun 2009 - 10:54 by ChrisCooke - put it back
This site is powered by the TWiki collaboration platformCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback
This Wiki uses Cookies