MPU Meeting Tuesday 21st February 2012

LCFG Server Refactoring

We are continuing to look at the rsync timeout issues on mousa. Chris has tried increasing the polling times on the various less important lcfg slave servers and has also increased the rsync I/O and connection timeouts but we are still seeing the problem. We don't see any errors on the other SL6 based slaves, even vole which is a VM on northern (in KB). We should upgrade trondra in the Forum to SL6 so we have an identical machine on a different site for comparisons. If we then only see the issue on mousa (which is in AT) we should investigate the network.

Simple KVM Service

There is now a simple rvirsh command which is provided via the bashdefenv package. Alastair has a newer version with slight improvements which should be available soon. The wiki documentation needs to be updated to use this new command.

SL6 Server Upgrades

Chris is preparing for the upgrade of the LCFG master by using a VM named tasman. To test the configuration he is using another VM, named coral, which is acting as an LCFG slave which uses the new machine as the master.

There are now only 37 LCFG schema packages missing from SL6, Chris will continue chasing people about this and removing this which are not needed.

Alastair will look at the ordershost side of the LCFG master server. It would be good to preserve the history in the client reports table.

We need to ensure we have up-to-date copies of all the data which is mastered on tobermory immediately before we start the reinstall.

We will need to give COs (and Sheila) lots of notice about this upgrade since it will not be possible to make any changes to LCFG profiles, headers or package lists for the duration of the reinstall.

Miscellaneous Development

  • lcfg-sleep : The exam lockdown behaviour has been fixed so that the lcfg-sleep package is not removed when a machine enters lockdown. This should avoid machines getting into a mess and not being able to sleep properly when they exit lockdown. Chris has also been doing a bit of thinking on how we could support sleep when there are X sessions running.

  • inf level headers : Alastair has fixed the inf level headers so that it is possible to login to an inf-layer machine again. This was related to changes which were made to how we enable nslcd in the dice-layer.

Operational

  • figgy : The FC test machine figgy is now working again and is ready to be moved to the KB machine room.

  • lochranza : The afs build machine lochranza has crashed, Stephen will investigate. We also need to fix the nagios monitoring so we know immediately when it is down.

This Week

  • Alastair
    • port dice-orders to SL6 (move into DICE svn, if not already there)
    • backup clientreports table
    • arrange figgy to go to KB
    • Discuss with George - how rack desktop servers in AT and IF
    • Start thinking about inventory project

  • Chris
    • SL6 defaults list
    • Continue work on tobermory
    • Complete ashkenazy replacement
    • Upgrade trondra to SL6
    • PD - local url shortener implementation (for not-a-service)

  • Stephen
    • Update the PXE installroot
    • PXE installroot - check documentation about moving link when building pxe root
    • Investigate lochranza
    • Continuing Theon work
    • Consider PD - concrete task

-- AlastairScobie - 21 Feb 2012

Topic revision: r5 - 24 Feb 2012 - 12:21:18 - 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