CloudServices/Meetings/2010-12-08: Difference between revisions

From MozillaWiki
Jump to navigation Jump to search
Line 41: Line 41:


* Question raised about memcache redundancy. If we lose one memcache, we lose the ability to give out JPAKE tokens from *all* memcache servers. '''AI:''' atoll and tarek to figure out plan.
* Question raised about memcache redundancy. If we lose one memcache, we lose the ability to give out JPAKE tokens from *all* memcache servers. '''AI:''' atoll and tarek to figure out plan.
* QA test plan for b7->b8 upgrade?
** https://spreadsheets.google.com/ccc?key=0AocUyLHteCtSdHNyLUVpUWpHdDlvWklwYmVaa2JMVXc&authkey=CJHPjucM&hl=en#gid=0


Answered in bugs, seems to be resolved
Answered in bugs, seems to be resolved

Revision as of 17:52, 8 December 2010

  • Time: Wednesday at 9:15 AM PST / 12:15 PM EST / 5:15 PM UTC.
  • Place: Mozilla HQ, 2J (JK)
  • Phone (US/Intl): 650 903 0800 x92 Conf: 8616#
  • Phone (Toronto): 416 848 3114 x92 Conf: 8616#
  • Phone (US): 800 707 2533 (pin 369) Conf: 8616#

Technical

Ops

  • scl2 status
  • lots of v1.5 investigation

Engineering

See yesterday's meeting notes for detailed status on projects.

  • J-PAKE: Had to get another NSS patch together
  • New crypto has landed (nightlies/1.6b1)
  • 1.6b2 shipped with compat bump for bookmarks.
  • 5 blockers left for b8
  • Async places API spec done, still waiting on resources from Mobile to accelerate.

Metrics

Product

  • 1.6b2 shipped to address issues in b1 and do the last compat changes (bookmarks only)

Roundtable

Notes and actions

  • Toby to turn atoll's email to an external facing blog post.

Fx 4 client

Fx 4 server

Other issues

  • Question raised about memcache redundancy. If we lose one memcache, we lose the ability to give out JPAKE tokens from *all* memcache servers. AI: atoll and tarek to figure out plan.

Answered in bugs, seems to be resolved

  • AI: MConnor to coordinate QA on JPAKE for both functional and failure modes in clustered memcache modes.
  • AI: tarek to do some load testing against staging and come up with numbers.
    • tarek: I ran them on dev (similar than stage): 1,250,000 Pake sessions per day with the current setup (1 pake session == 30 requests)
    • need to bench on prod now
  • AI: rnewman to get some numbers on how much better the perf is with the simplified crypto work.