MPU Meeting Tuesday 15th September 2009

Power Management Project

The sleep component is now running on all the HP DC7900 machines in the student labs.


The dates for the change to the new system have changed. Alastair now plans to do this on Tuesday 22nd September.

AFS Component

As part of the pandemic planning work Stephen has been working on preparing for the roll-out of the new openafs component to the DB and file servers.


Not much happened. Chris will put together a plan for how the new TIBS LCFG component can be brought into service.

LCFG Server Refactoring

Nothing happened

Miscellaneous Development

release scripts
Some changes were made to the release scripts to make them more standard. Also a new script was added for creating branches and this was documented on the ReleaseManagementProcedures page.

Pandemic Planning

Whilst writing up the details of how to restore the LCFG master if it dies Stephen came up against a few awkward issues related to retrieving data from the mirror servers.

He suggested that the rmirror component could be improved with the addition of a spanning map which holds an inventory of all the data being backed up. This could work in a similar way to the standard inventory component. The addition of a simple CGI script would make it easy to view and search the information. We will bring up this suggestion at the Technical Meeting in October. For now Stephen will make a list of all the MPU data being mirrored so we can find it easily in a hurry.

Stephen also noted there are no clear instructions on the best way to get data back whilst preserving modes and permissions. This can be a pain when there is a lot of data involved.

It was agreed that we should look into having a virtual machine at KB which acts as a hot spare for the LCFG master to avoid bootstrapping issues and to make the replacement as speedy as possible.

The addition of ethernet bonding to all MPU servers is now completed with the switch to the new header.


New forum-server-room header
The new new forum-server-room.h header was added to all the relevant MPU machines and the nut component started. A question was raised about the best approach for virtualised servers.

bakerloo has been switched back to the stable release.

New R710 machines
Once the KB machine room is ready we will get a couple of R710 machines up there so we can deploy some machines off-site (like a hotspare for the LCFG master).

This Week

Alastair will:

  • research pandemic interests
  • write pandemic howtos
  • finish Nagios testing in etherbond.h
  • reboot central

Chris will:

  • research pandemic interests
  • TIBS adoption plan

Stephen will:

  • research pandemic interests
  • document disaster recovery for LCFG master DONE
  • Make a list of MPU data backups DONE
  • update LCFG server documentation
  • test AFS component database servers

-- StephenQuinney - 18 Sep 2009

Topic revision: r1 - 18 Sep 2009 - 09:13:38 - StephenQuinney
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