Meeting Notes for 26 August 2009

Systems

• A serious database MySQL failure occurred 18 August
  • Argonne MCS had never been informed of said database
  • Backups were not properly done
  • Data was mostly - but not wholly - recovered
  • Ken Raffenetti at MCS is setting up proper database backups
• www18 disk configuration is seriously screwy and prevented the server from booting itself
  • The original RAID configuration on www18 was really, really weird and at the time of the upgrade to 8.04 something strange happened with it. Fixed, but still a looming issue
• data2 upgrade to 6x2TB RAID6 array (8TB effective space) is proposed to replace our
• Longer-term consideration of server consolidation
• Switch to GigE network to cut down our network bottleneck
• Give a read- and lock-only backup account for the Postgres and MySQL databases

Forums

• No sticky posts (only sticky threads)
• Set pagination? Newest-post-first? Jump-to-newest-post mode on by default?
• Edit timeout might be too short?
• Eventually switch to the Wiki to handle teacher workshop content

Wikis

• Working on finalizing recovery of missing wiki content and the old tutorial
• /teacher/$elab wiki's configuration was screwy (and now fixed; relevant LocalSettings.php file in SVN)
• /library lost some metadata tracking tables and revisions - looking into this.
  • Some LIGO glossary data is gone and needs to be recreated
  • Image table pointers are missing - trying to recover

All e-Labs

• Logbook is unintuitive and really old code and probably needs to be worked on
• Major bughunting, security-hardening, etc. initiative in progress

LIGO

• Eric's gotten GDS working on his laptop - needed to read new LIGO frame files

Cosmic

• Issue with the old test still becoming active - working on it
• Weird double-insert issue with the legacy administrative group creation tool
• Data Blessing issues - TomJ is working on an Excel workbook tool for data calibration.
• Specs are needed for the code changes for the new blessed data uploads.
  • Default search then looks for user's own (or school's own?) data and all fully blessed.
  • Needs some more thought

CMS

• Login screen for OGRE? Why is there an ID? Do we need a password?
• Integration?
• Keep our own OGRE build and occasionally synchronize with the upstream OGRE source tree?
• Transparent proxy to Notre Dame's hosted OGRE?
• Cool Science content is needed
• Link to blogs for particle physicists?