Wikimedia Platform Engineering/MediaWiki Core Team/Check-ins/20140407

who: Nik!, Greg!, Bryan!!, Brad?!, RobLa, Ori, Chris, Aaron, Dan, Tim regrets: Antoine

Bug escalation

edit

https://bugzilla.wikimedia.org/show_bug.cgi?id=54203

  • CSteipp and Dan will work out the exact ask, and email it


  • l10nupdate
    • add snapshotting to the l10nupdate script for today’s
    • run scap again today, see if it changes things? (but that’d be weird…)
    • Bryan to check in with Tim/Greg at end of day, hand off to Tim if needed


Quarterly Review

edit

Revision storage:

  • Sean doesn’t think the revision table is currently a performance bottleneck. “The true pain is in special page queries, mainly anything under IndexPager: they could all use a rethink.” He is interested in a design stage comparing TokuDB with Cassandra: “Tokudb engine is almost as exciting as cassandra. revision, logging, the *links tables, anything with more than a couple indexes, could all be candidates for tokudb”
edit
  • Deployed to most of the wikis last week. Performance just fine, but none of them were big.
  • Upgrading to Elasticsearch 1.1.0 Tuesday.
    • We’ll use this opportunity to test out the new git-fat based fat artifact deployment for Elasticsearch plugins. We’re the guinea pigs here.
    • Going to do another round of performance testing after the Elasticsearch cluster has stabilized after the upgrade. Should be better than last time but will definitely find more to improve.
    • We have tons (5ish) waiting on this release so we’ll get started on those as soon as it is in production
  • Highlighter
    • Deployed to beta but not in use (waiting on the review from Chad)
    • Released to the public (on maven central)
    • Moved to Gerrit
    • Still waiting on a review from Elasticsearch folk

Performance / HHVM

edit


Deployment tooling / RelEng

edit
  • grumble l10nupdate breakage


SecurePoll

edit
  • Nothing noteworthy to report from Brad.


ContactForm for trademark

edit

Security

edit

Hiring

edit