CloudServices/Meetings/2009-09-14

Agenda:

0.7 goals:

• Don't keel over
• Don't turn away users
• Keep it simple, especially on fennec
• And oh yeah - something about 1.0 Beta comes to mind

• 0.7 status - where are we and what's needed
  • Partial sync [mardak]
  • Backoff [mconnor]
  • Handling race conditions - need an owner for this [mardak/mconnor]
  • about:weave [ragavan on behalf of dan]
    • Need help on fennec
  • Server changes? [toby]
  • Pre3 [all]
    • Soft freeze everything except about:weave
  • Disaster planning walk through - what happens when load peaks? [zandr/toby]

• Service scalability project https://wiki.mozilla.org/Labs/Weave/Service/Scaling/Projects/Aragog [zandr]

Notes:

• Should we set up a different cluster just for fennec users?
• Partial sync
  • We need some way to show users that all of their data isn't there yet.

AI: Mardak will mock up the UI.

• Backoff
  • Going out for review tonight. Will be part of pre3.
  • AI: We need some server side header support.
  • Need to plan the server side stuff, perhaps roll this out a part of pre3?
• Handling race conditions
  • With backoff stuff in, this may perhaps not be as needed?
• about:weave
  • Dan's patch is pending review. Looks good.
• Server changes:
  • List of modified timestamps - can be memcached. This is live on the dev servers.
  • Collection counts - checked in. Not live on dev server yet.
  • Backoff stuff - This is live on the dev servers.
  • AI: We need to get QA involved somewhere.
• Disaster plan:
  • Backoff is our biggest bet.
  • 9K concurrent 0.5 users at peak, 10 AM Pacific.
• Ordering extra hardware:
  • If we go based on our current load profile and order similar hardware, we are looking at about 150K. Will check with Chris on options.