Our remote results are still noisier than I would like. I feel there are
unknown factors that are affecting our remote results. It is difficult
to analyze root causes of the variability in the measurements we are
getting due to our networking situation:
1) We are using my wifi network which is subject to congestion and
interference between the phones.
2) We are using ipfw+dummynet to shape the network for the phones in
Autophone. I don't have measurements of the behavior of the data without
ipfw+dummynet and can't compare its effect on our remote measurements.
For #1, we have the suggestion to use ethernet over usb in bug 969518
[1] but I think this might be problematic especially on older versions
of android. This might take a bit of time to implement and I would
rather not gate on it.
For #2, I would like to temporarily stop using ipfw+dummynet so that I
can do a before/after comparison.
I feel the remote tests are currently noisy enough that their utility is
limited and that this change would not negatively affect us. If you
really truly miss the ipfw+dummynet results, I can re-enable it after a
few days.
Proposal: Stop using ipfw+dummynet beginning with 2014-02-08 builds.
Revisit in a few days to see if we should turn ipfw+dummynet back on
before we have the ethernet over usb solution in hand.
/bc
[1] <https://bugzilla.mozilla.org/show_bug.cgi?id=969518>
_______________________________________________
mobile-firefox-dev mailing list
mobile-firefox-dev@mozilla.org
https://mail.mozilla.org/listinfo/mobile-firefox-dev
