MPU Meeting Tuesday 17th December 2013

Inventory

Alastair intends to write a comprehensive snapshot of the current state of the project, then stall the project.

Virtual DICE

No activity.

Client Refactoring

The parsing code can read in an XML profile and generate a tree. Given the context information it can give the correct value of a resource. It can now also write it out to the DBM file and read the current state from the DBM file. So, it's now possible to compare the current state of the profile with the new state.

Until now the definitive source for the current state of a machine's profile has been the DBM file on that machine. However with the new code the definitive source will be what the client holds in memory. It will always write a backup copy to the DBM file, so the latter will always be an accurate representation of the current state; but the change means that there will no longer be a need to read in the DBM file before doing a comparison. The DBM file is used directly by the ngeneric component - another reason why it will always be kept accurate and up to date.

Stephen has written a tool which, when given two profile objects, will give the differences in resources between the two profiles.

To test the new code he plans to build DBM from every XML profile we have and compare these with the DBMs generated by the existing v3 code, and look for any differences.

This leads naturally on to a new development project - LCFG::Profile, to produce a standard way of reading profiles.

Links:

LCFG Git Feasibility

Chris intends to write a comprehensive snapshot of the current state of the project, then stall the project.

He has been looking into how to split the existing lcfg configuration repository into separate dice and lcfg repositories. The most promising approach he's found so far is to split the subversion repository into two separate subversion repositories using svnadmin and svndumpfilter.

Stephen and Alastair made some interesting suggestions:

  • Don't try to preserve the current shape of the repository. Instead try to find something which suits git. (Stephen)
  • Propose various models. Don't just choose one. (Alastair)
  • Prototype it. Hack together some sort of example repository; plug in a test LCFG server; build profiles. That will allow you to test changes, try ways of doing releases, etc. (Stephen)
  • With separate dice and lcfg repositories, look into the problem of transferring a chunk of configuration from one to the other. At the moment this can be done in one atomic transaction without triggering a change in XML profiles. Could this be achieved in the new setup, avoiding the problem of (for instance) unnecessary temporary changes to spanning maps? (Alastair)

Miscellaneous Development

Stephen and Alastair have prepared the way for a project to port LCFG to SL7. As well as the normal LCFG porting activity this project will have the considerable task of adapting LCFG to support systemd and to configure networking using NetworkManager. See the links on the Scientific Linux 7 LCFG port project home page for more details of systemd, NetworkManager and the SL7 port in general.

We have a VM running the current RHEL 7 beta.

Operational

  • The appearance of apparently back-ported SL package updates makes it look as if SL 6.5 is about to hit us. To minimise support hassle we plan to support 6.5 as soon as possible.
  • Alastair has upgraded devproj to Drupal 7.24.
  • Stephen rolled out a critical update to PHP.

Next Meeting

The next MPU meeting will be on Tuesday 7 January 2014 starting at 2.15pm.

This Week

  • Alastair
    • Start Inventory project diary
    • Inventory project
      • Submit bug/enh to App::Cmd author wrt option to die on unspecified options
      • Pester George about location API
      • Publish revised schema (and document)
      • Continue fleshing out processes (by end of 2013)
    • Order a spare 600GB disk for waterloo (hot spare)
    • Ask George - what does the TXretransmit value mean for switch connections? (Can't see this using rfe -xl )
    • Consider how to make metropolitan usable by users
      • Decide whether we do this now or not
      • ISOs
      • minimal docs (mostly manual)
      • they'll use virt-manager, but not create machines or change config
    • circulate table of LCFG bugs
    • Consider activities list
    • Look at gnome issue with NX - try inftest account
    • Finish off RT tickets
    • Read LCFG XML documentation
    • Look to see if there is gconf option to remove "shutdown machine" from menu item (quick look)

  • Chris
    • Work on GIT project - produce brain dump on wiki
    • Tidy up RT tickets as per discussion 17/12
    • Ask Graham how easy to implement an rfe command to archive an LCFG profile - circulate Graham's response to MPU

  • Stephen
    • LCFG client refactor stage 1
      • report
      • schedule debrief meeting for new year
    • LCFG client refactor stage 2
      • continue working on mock up
    • Consider dhcpd component changes
    • Consider activities list
    • Produce wiki page listing what needs changing when archiving an OS platform
    • Add LCFG proposals to Ops meeting discussion list

  • Carol
-- AlastairScobie - 17 Dec 2013
Topic revision: r9 - 07 Jan 2014 - 13:41:32 - AlastairScobie
 
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