Project Development Meeting

(JCMB Room 2511)- Minutes 2007-09-05

Present: Paul Anderson, Toby Blake, Neil Brown, Tim Colles, Chris Cooke, Ken Dawson (convener and minutes), Graham Dutton, Bill Hewitt, Stephen Quinney, Iain Rae, George Ross, Alastair Scobie, Craig Strachan, Simon Wilkinson

Apologies: Ian had sent his apologies.

1. Minutes of last meeting

These were accepted.

2. Matters Arising

Seventeen out of the twenty seven computing staff had now moved their mail service to staffmail.

Paul has now moved but Simon will not be able to move for another month or so. Neil reported that the Default quota had now been increased to 2 GB. However this was a limit that could not be exceeded until IS introduce a mail server with a later version of Cyrus software.

Richard and John have also still to move. [Action John, Richard, Simon]

  • Introduce a Production FC5 Desktop Iain has produced the final version of the report. He will link to it from within the milestone for producing the report. [Action Iain]

All other actions from the last meeting have been completed.

3. Approved Projects

  • AFS file system Other higher priority tasks have impacted on the achievement of this last milestone. Craig will discuss with Alison what a suitable revised date should be and propose it to CEG. [Action Craig]

  • Production Cosign Service The final step of including spnego changes to firefox config for all DICE users is almost there. Firefox 2 on FC6 already has the change. Iain hopes to make the change for Firefox 1.5 on FC5 some time next week.

  • Production Condor Service Iain has had to revisit the way condor jobs are affecting lab (and other condor pool) machines. When condor jobs take up too much memory (or more correctly when the amount of free real memory becomes too small) slapd is frequently seriously affected, either by being killed off by the kernel or the Berkely Database back end running out of some resource.

Iain will reset the next milestone to being the 'all labs' milestone, for 'Deploy in all labs and volunteer user desktops' and reset the date for this and subsequent milestones that still need to be achieved. [Action Iain]

  • OpenLDAP Replication and Server Configuration Since the project was suspended new candidate technologies had become available that now need to be investigated. We are fairly confident that we have already identified the most appropriate solution for master to slave LDAP replication and it would be best to implement this now. It was agreed that the rest of the work of the original project should be split off and formed into a separate new project. If this new project needs approval to commence before the next Development Meeting then Toby should make the request via email. [Action Toby]

  • Gridengine Configuration Review Iain asked for this project to be moved to the stalled state as any further work was dependent on the implementation of a working GPFS. It was agreed that the status should be so changed. [Action Iain]

The following project was signed off by the Development Meeting and goes to CEG for ratification:

4. New project proposals

Stephen made the case for the reimplementation of an LCFG buildtools framework. The current system was too tightly tied into using a specific version control system (CVS) and had a test framework that was both poorly documented and rather difficult to use. Goals of the new system would include being independent of the use of any particular version control system (including not using one at all), have a more usable test framework and the ability to build software on several architectures using one command rather than having to build the software repeatedly, once on each architecture. The new project had the full backing of the Managed Platform Unit.

It was agreed that this project should move to the evaluation phase.

5. Technical presentation and discussion of projects

  • Tim talked briefly about the proposal for a Service Catalogue. In the ensuing discussion several helpful points were made. Use of standard elements in constructing the access/authentication to a service should be recorded in as simple a manner as possible. Standard elements should be recorded in a machine-readable way (possibly using XML). Non-standard methods and schemes should be highlighted and recorded in a way that allowed free text. Each service could be described in a single file and LCFG profiles of machines providing that service should be tagged to refer to this service description. We could aim to start off with a fairly basic set of data fields and slowly increase the richness of the data held in an incremental way.

  • Simon gave a brief presentation (PDF) on the new apacheconf component.

6. AOB

The next meeting will be on October 3rd in the Buccleuch Place Seminar Room.

-- KenDawson - 05 Sep 2007

Topic revision: r3 - 02 Oct 2007 - 15:56:39 - KenDawson
 
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