DeveloperServices/TeamMeetings/2015-01-06

« previous meeting — index – next week » ^create?

Meeting Info

• 2.00 pm Pacific Time Wednesday, Jan 7 <= **One time change**
• Laura's vidyo room
• Backchannel in #vcs

Hot items

• On Call Handoff!!!

Last week

• bkero
  • Started work on github: Created Github Pilot Etherpad
  • Requested Github Enterprise AWS trial, started setting up
  • Code reviews for various hg scripts
  • Created/deployed git 2.0 to staging for testing
  • Began hgweb.m.o CI work
• fubar
  • TLS and HSTS for DXR
  • BMO production update script love and fixes
  • Treeherder newrelic update and fixes; everything finally reporting correctly
  • vcs-sync walk through
  • SVN user permission fixes (what year is it?!?)
• gps
  • pushlog rewrite
    • ctrl+c during push no longer results in inconsistent database state
    • slaves pull pushlog data incrementally, leading to faster replication and pushes
    • *hg verifypushlog* now exists. extension has revset and template support (bug 1116886)
    • only 1 regression on deploy, and it only impacted json-pushes endpoint of empty repos (part of the legacy pushlog code)
  • hgweb templating work
    • attempting to bring theme on hg.mozilla.org up to date so more useful
    • https://reviewboard.mozilla.org/r/1887/
  • Firefox Sphinx docs work
    • http://gecko.readthedocs.org/en/latest/ now works \o/
    • http://gregoryszorc.com/blog/2014/12/30/firefox-source-documentation-versus-mdn/ led to renewed "better docs on MDN" discussions (bug 920314)
  • version-control-tools is now tracking BMO master branch again (bug 1118650)
    • we were pinned on a commit from October due to BMO-in-Docker bugs. Upstream work to support Docker has flushed these out!
  • firefoxtree hg extension features (bug 1116869, bug 1116861)
  • Trying to prod Review Board into sensible packaging (https://groups.google.com/d/msg/reviewboard-dev/5_H9jULhYUs/ZNEsJOh1PWAJ)
• hwine
  • start cross training on vcs-sync
  • close old bugs
  • start github planning
• laura
  • Trello: https://trello.com/b/KwwYSXE1/release-engineering-status-board
  • If you are not on there, get an account and I will add you
  • Individual deliverables due this week!
  • git/github discussions with Hal

Planned for this week

• bkero
  • Finish setting up github enterprise AWS instance(s), start populating it with Real Data (TM)
  • hg.m.o CI - Get hgweb side created, start testing against it
• fubar
  • BMO push duty
  • AWS/terraform/docker learnin' and testin'
• gps
  • Land and deploy hgtemplates updates (waiting review)
  • Browser-based (presumably Selenium) test infrastructure for MozReview
  • Bundle-based repo work (try scalability)
  • Try to finish some Mercurial patches in time for 3.3 release
    • JSON API in hgweb (would allow us to eventually kill a lot of custom code)
    • Make hgweb templating more powerful so we can do more powerful things with the web interface
    • Make the transaction API tolerable (so pushlog works more robustly)
• hwine
  • more github planning and meetings
  • (release build week may take work)
• laura
  • Finalize q1 things for everyone
  • treeherder/tbpl parity

Other business

• mailing lists -- too many?
  • infra-devservices@m.c <= will kill unless anyone hollers
  • developer-services@m.c <= Hal's list for "team" & covers all operational systems
  • version-control-dev@m.c <= Greg's list for DVCS work
• gps priorities
  • Push times to try have decreased dramatically due to peripheral optimizations that are not bundled-based repos. Remaining long pole in push times is synchronous slave replication. This is a problem for all repos. We also have very annoying replication bugs. At this point, I think Try is under control and replication lag/bugs are our biggest performance and business continuity concern. I think bundle-based repos should be put on the backburner until replication is better.

PTOs, etc

• fubar - 16-26 Jan
• bkero - 10-16 Jan - Linux.conf.au presentation

Links

Goals