MPU Meeting Tuesday 2nd November 2010

LCFG Server Refactoring

Stalled.

Software Build Farm

Stephen reported that documenting the database layer was now finished. He is now looking at the daemonized processes. The build daemons now run without actually executing mock. There are issues with Perl signal handling, timeouts etc and this requires care to make it robust.

F13

This is now closed.

Replacement for VMware Server

Alastair has created a homepage for the project. This needs to be published and discussed. It may be that we want two services. Since 'live migration' is hard and would increase the project time then we may be able to use the IS service (assuming it is suitable) for those services that require it. Replacement of our own VMware service should give us better manageability and peformance for local virtual servers. We should evaluate IS's service.

Review Project List and Prioritise.

There was a long discussion on the current and pending projects for the next project prioritisation round.

• RHEL 6 based distribution : it was agreed that this project would be a main priority.
• Build Farm : should be finished by Christmas, thus allowing us to revive...
• LCFG server refactoring : as a priority for T1-2011
• Virtualisation : even if we go the IS route for some VMs this project still needs effort.
• Install scripts for self-managed : evaluate requirements, was down for T3-2010.
• Fstab enhancements: raise the priority of this project. Potential joint project with Kenny at IS. Use Perl not C.
• Evaluate non-RHEL based Linux distributions : no time.
• LCFG documentation technology : no time
• Logserver enhacements : schedule after the LCFG server refactoring.
• Automated upstream package monitoring : Perdita is keen to see something like a web front end so we can see what packages are installed.
• Virtualised DICE image : (Takeaway DICE) this will involve moving around large quantities of data.
• Package submission enhacements : Chris is doing this ?

It was pointed out that 'user visible' projects should be given a high priority if possible and

Miscellaneous Development

Patch to pkgsubmit

to prevent submitting a file twice

Operational

Panos's request

no action

Moving sauce to KB

this has not yet been done as it is awaiting the technicians

Architecture document

started

Moving MPU VMs

moving guests from bakerloo, started

IPMI power

memory problem prevented further action ?

AT machine moves

not done yet

Nagios mute

done

F13 KDM problem

fixed - it was old dotfiles causing problems for Lindsay (and others)

smartd problems

An application called palimpest running on F13 is erroneously reporting smartd errors. Agree to pull from DICE level.

Perl training

Alastair reported that Liz says there is no money available for this training course.

Devel bucket

to disappear. Other units have been informed so we can go ahead and do this.

Updating fstab defaults

Kenny has suggest that now that hd has gone and disks are SCSI we should look into fixing the defaults. Perhaps wait until RHEL6. Add to the activities list.

Kernel bug

Stephen had circulated a message about a kernel bug. RHEL fix awaited, fix when it comes.

Memory problem

R710s : Chris has phoned Dell

F12 package list

removed, but 3 hosts still complaining :

Powerdown

the powerdown revealed that refreshpkgs uses AFS but starts after it in the init sequence. Stephen to fix.

This Week

Alastair will:

• VM server replacement project
• Pkgsubmit patch

Chris will:

• VM server replacement project
• IPMU on Northern and Piccadilly
• remove gnome-disk
• RaT submit project

Gordon will:

• Review LCFG bugs

Stephen will:

• Disk space on telford and mirror
• Refresh pkgs (as minuted)
• Build farm project

-- GordonReid - 15 Nov 2010