MPU Meeting Tuesday 4th October 2011

AFS automation project

No further activity.

LCFG Server Refactoring

Awaiting confirmation of Simon's availability.

Wake on LAN

There was a broad discussion of some of the issues which will affect this project. Only authorised users should be able to wake a sleeping machine but e.g. DHCP has no idea of who a machine is allocated to. James has a wake-on-lan client component in IS, which could be adapted to take out EASE/AD things and in any case the schema should be generic. A mapping of MAC addresses to the people machines are assigned should be in the inventory. New self-managed machines are now being given profiles with placeholder headers but older machines may not have this.

Install Scripts

Report due, but won't be ready for Development meeting this week. There is no obvious clear course of action as a follow-on from this project. Report to be done before wake-on-lan work.

Submit

Issues are emerging in use which indicate possible improvements. It was felt that MPU time for this project had been used up and any further enhancements should be left to RaT unit.

Simple KVM Service

Alastair has being working on scripts to create and delete machines. There is an issue with time shifts after clients are suspended/resumed, as the server only provides 'ticks' not absolute time. Practice elsewhere might be to live-migrate VMs before rebooting the host server? There was discussion about changing the boot order on creation, and the limitations of the Simple XML perl module. libxml would be better.

MPU Server upgrades

Stephen is planning the SL6 upgrade for rpm cache and pxe servers (split and schiff). He will use prague for testing and add it in to the cache.pkgs pool for load balancing while he does the other upgrades. There are changes to squid e.g. to netmasks. The lack of the perl AFS module on 64 bit stalls things. We want AFS 1.6 everywhere as soon as possible. Alastair will look at waklog.

Miscellaneous Development

Teaching packages

Stephen has added an override for teaching package additions - live-sl6-overrides.rpms on all SL6 stable machines. He will do this for SL5 machines too.

fstab patches

Alastair has received a patch from Mike Gordon for LVM partition types. The revamp of the fstab component should roll-in lvm stuff eventually. LVM is handy for virtual machines. Alastair will also respond to Matthew's question about the installroot.

Network component

there is new support for KVM bridges and the addition of a delay time to fix PXE. The STP stuff has been disabled, due to the loop-detection issue.

Firmware

Alastair has been gathering bios upgrades in the unit group space, e.g for idac cards and raid controllers. Need to document how to find out what current firmware is. There are utilities available for e.g finding out about raid arrays. The HS200 still has an issue where the raid controller can't see things with megacli.

Operational

F13 build host reboot

vmware can't boot the 32 bit kernel. Needed to downgrade the kernel and reinstall the machine.

Circle

reported faulty DIMM. Replaced it and firmware upgraded, but it's still moaning.

Bakerloo

is now back up and the battery for it's raid recharged.

Activities List

The list was reviewed and a number of items noted as completed. It was agreed to set up a separate meeting to discuss the 3 year plans for server purchases. (Wed 12th at 14.00).

DONM

Next week's meeting will be tuesday at 11.00.

This Week

• Alastair
  • KVM project
  • SL6 server - Look at waklog
  • Respond to Matthew re LVM in installroot
  • Chase George re solution for DNS problem (wrt small servers)
  • Look at MPUBackupSummary
  • Document BIOS settings for USU
  • Think about 3 yr equipment spending profile

• Chris
  • Finish install scripts report
  • Think about 3 yr equipment spending profile
  • Start WakeonLAN project
  • PDP

• Gordon
  • Student Lab availability

• Stephen
  • SL6 server - perl-afs and openafs-1.6.0
  • SL6 server - upgrade split
  • Think about 3 yr equipment spending profile
  • PDP - python learning for ISS DB work

-- AlastairScobie - 04 Oct 2011