MPU Meeting Tuesday 2nd March 2010

AFS Component

All of the AFS fileservers and database servers are now using the new component. Nearly all clients have now switched. Some servers haven't yet switched because they haven't rebooted yet. Stephen's work on this project is now pretty much complete. It's possible that a few bug fixes may be needed but that's it. (Congratulations Stephen!)

LCFG Server Refactoring

Stephen has finished porting the code to the perl Module::Build system. Next he'll add git support to BuildTools - just basic support to enable it to make tarballs, RPMs and so on from a git repository.

Installroot

Alastair has packaged up the code to build the new initrd image and made the necessary changes to buildinstallroot.

The next stage will be to tackle the code in the installroot. Alastair will replace rc_install with new code.

F12

Chris has been blogging about his progress on the F12 project.

Most of the core is now done. A few tricky dependency issues have come to light:

  • Don't believe all that you read about dependencies in the comments in the LCFG package lists; these comments may be out of date.
  • Perl software shouldn't normally need its perl module dependencies mentioned in its specfile (though may in more complex cases as the code that spots dependencies is pretty simple-minded) but despite this such dependencies will still need to be mentioned in package lists.
  • Unless they're build time dependencies ("BuildRequires" in the specfile), in which case they just need to be there at build time. Such dependencies can for instance be added to the live/buildhost.h header.

Boot

Alastair gave us a live demo of an F12 machine booting using the boot component! He's replaced "rc" events with "lcfg" events. The test machine doesn't yet also start other LCFG components.

When we ported to SL5 we didn't go through the run level numbers to make sure they were still correct, we just adopted the ones from the previous OS. Some of the numbers aren't right any more. Alastair's going to go through the numbers and make sure they're right.

This week he expects to finish and package up the results and add to headers.

Miscellaneous Development

Package Building Submission System
or Stephen's Amazing Build-O-Matic. He's put together a prototype system which accepts job submissions of source RPMs with simple instructions, validates those jobs then builds the software on all applicable platform/architecture combinations, farming it out to "Mock" on relevant buildhosts. This should be really useful, not least in the F12 port. He's got about another half day of work to put into it before it's ready for use.

Operational

Unusual monitor & graphics card configurations
These aren't now worth documenting until we're using F12, which may deal with such things automatically anyway, so we'll revisit this action then.

Mirror progress
Stephen found that he was after all able to change the ownership of files in AFS, but only once he'd "kinit"ted to his admin principal. He's therefore now been able to finish setting up our F12 packages mirror. He'll shortly add SL5 and EPEL packages mirrors too. All of them will be found on our "sites" tree at /afs/inf.ed.ac.uk/pkgs/sites.

RAID for VMware servers
Alastair has finished his RAID config testing. Once bpbeast has been released from its current duties Alastair will deploy it in two halves for two VMware servers: fourteen disks will be used for central and fourteen for bakerloo.

Replacement for VMware Server
Finding a replacement will be a substantial task, so it's been moved from here to our pending project list. It now has a devproj entry (154).

AT Server Move Complete
Stephen has cabled up mousa and split with 4m network cables. This completes our server move from FH to AT.

Server Update
Stephen's bug fixes for the stable lcfg-server have been tested by Kenny and Stephen. The new version successfully compiles all the profiles compiled by the currently deployed version. The new profiles haven't yet been tested to ensure that they're as correct as before.

lcfg-boot
Stephen has fixed the problem described last time.

Server reboots
Chris and Carol have rebooted a bunch of MPU servers. Stephen will reboot whatever's left in the next few days.

lcfg-logserver bug
Stephen needs to test his solution to Bug 227 on a MacOS system. Alastair's giving him the MPU Mac laptop.

This Week

Alastair will:

  • Merge Panos' changes into the network component
  • Look at MPU disaster recovery
  • Consider periodic file configures
  • Think about MPU logging requirements
  • Talk to George about routing problems
  • Think about F12 package buckets
  • installroot work

Chris will:

  • Look at MPU disaster recovery
  • Consider periodic file configures
  • Think about MPU logging requirements
  • F12

Stephen will:

  • Look at MPU disaster recovery
  • Consider periodic file configures
  • Think about MPU logging requirements
  • Server refactoring
  • Finalise lcfg-server changes

-- ChrisCooke - 4 March 2010


This topic: DICE > WebHome > ManagedPlatformUnit > MPUActivitiesList > MPUnitMeeting20100302
Topic revision: r2 - 09 Mar 2010 - 13:34:32 - 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