Operational Meeting

Craig, Stephen

Toby pointed out that his document on the procedures to be followed for the management of the server certificates was on the inf-unit documentation pages, not the wiki, for bootstrapping reasons!

Actions completed:

• Neil has added guidelines on AFS and ssh to the "top 10".
• Toby has created guidelines on how to manually add signed server certificates when reinstalling

Actions added:

• Chris will look into wake-on-keypress for sleeping machines

Report from Computing Executive Group

No report. It was noted that the published minutes were a bit out of date. George will upload the most recent ones (since done).

Reports from units.

• Infrastructure.

  4. Cosign v3 upgrade
     • Iain asked about the St. Andrews cosign-protected pages. Toby has been in touch with them directly.
     • There are some LCFG spanning map issues. See the wiki page for details.
     • At present the "allow v2" flag is set for all services at the server end. At some point we'll be turning this off, so that service managers will have to either upgrade or set it explicitly themselves. See the post-meeting note in the inf-unit report.

• Managed Platform.

  • Iain asked whether wake-on-keypress was possible, as it wasn't always convenient to press the power button. Chris will look into it.

• Research and Teaching.

  2. ProofGeneral Wiki

     Graham observed that the ProofGeneral wiki seemed to be a spam magnet, and suggested that the wiki managers might want to keep an eye on it.

  • Iain observed that he'd be holding off with firefox-3 for a few days so as to be able to pick up 3.5.

• Services.

  • There had been an evo fault which was causing home-directory problems this morning. A disc had failed, but for some reason the spare hadn't taken over and partitions had gone read-only. Craig was fixing things up, but it was taking a while as fscks and raid-rebuilds competed for resources.

    ... and thanks to Graham for a working DICE account to allow us to use the data projector for this Operational meeting!

  • alexandria (the TiBS machine) had hung a couple of times for no obvious reason. The second time Craig had recovered it remotely using the power-management system; instructions for doing this have since been put on the services-unit wiki pages. Other than this, the backups were now back to normal following the recent problems, but there were still a few things to be thought about and tidied up.
  • We expect to hit 1/4-million CUPS jobs very soon! There was some discussion about job names in the queue listings possibly revealing information that perhaps shouldn't be, but no definite conclusion was reached.

• User Support.

  No report.

Topics for discussion

• Ending support for SSH public keys

  This was held over for another meeting with a wider range of attendees.

AOCB

None.

The next Operational meeting will be on Wednesday 22nd July in IF-4.31.