TestMtg, 2008-01-07

From OLPC
Jump to navigation Jump to search

Attending: Jim, Kim, Chih-yu, Daniel, Michail

Agenda Items:

Automated Testing - Tinderbox

  • Try to get tinderbox back up and running

Activity Testing - Goal is to push this into the community for testing

OLPC testing (things that can only be tested at 1CC)

  • Connectivity
  • Performance
  • Scaling
  • New features/functionality

Testing thoughts for Update1:

  1. Need to resurrect the list of B2 exhaustive test spreadsheet to get into major release
  2. Need to augment one-hour test to include all the new features
  3. Need to add known bugs to the test cases: ie. when you change X, you need to test scroll pad
  4. Need to add scaling and performance tests
  5. Need to add school server testing with Laptops
    • Signed OS images can be reinstalled from the school server
    • Backup from XO to school server; restore
    • olpc-update should work with the school server
    • local jabber
  6. Olpc-update on all three supported encryption infrastructure AP.
  7. New wireless driver testing

Upcoming things:

  • Security fix in X is coming; will we have to take it in Update1?
  • Q2D08, too many bug fixes for the ship2.2 patch - would be good for Update1
  • EC code fixes for suspend and resume could go into Update1
  • Wireless firmware, p49


  • Is it going to be possible to turn OFF the automated update - should this feature be in ship2.2?
  • Can Dennis help with 656 usb missing file?


Chih-yu:

  • One-hour smoke test took about 2 days - due to network access and some journal interaction.
  • One thought is that you should reboot just before you start 656 testing... this shouldn't be as bad in joyride/Update1
  • If there is still a very slow sugar frame bug; it needs to be logged in trac
  • If there is still memory hogging after 2 days of no reboots; that is a bug
  • New content bundle still can't be installed on joyride, as of 2008-01-07 (joyride-1489). See #5840 -- Chihyu 15:05, 7 January 2008 (EST)


Michail:

  • Wireless driver was entirely rewritten and will need a lot of testing.
    • Perhaps this big a rewrite is too risky for Update1
    • p49 put probes back in for ad-hoc support
  • 5.110.22.p0 is currently under test
    • Some timing related issues, such as retries
    • Remove some excessive debug events

Control panel feature requests:

  • Be able to turn off all probe requests
  • mDNS doesn't get repeated by the mesh in a classroom environment (dense/classroom scenario) for specific applications