Infrastructure load for April 2012

NOTE: In April, we lost 20% (6 of 30 days) of metrics data during the switchover of the backend databases from sjc1 to scl3 data center. Despite missing 20% of the entire month’s data, we still recorded more jobs in April then entire November2011 or entire December2011.

  • #checkins-per-month: April looks like a drop, with “only” 3,327 checkins, but this drop is under-reporting because of the 20% lost data. For comparison, previous records are: March2012 (4,508 checkins), February2012 (4,027 checkins), January2012 (3,962 checkins), December2011 (3,262 checkins), and November2011 (3,209 checkins).
  • #checkins-per-day: We set three new records in April: 280 checkins per-day on 25-apr-2012, along with 278 checkins-per-day on 26-apr-2012, and 246 checkins-per-day on 24-apr-2012.
  • #checkins-per-hour: We peaked at 7.1 checkins-per-hour, which is lower then usual, but to be expected given the lost 20% of data.

mozilla-inbound, fx-team:
mozilla-inbound continues to be heavily used as an integration branch, with 22% of all checkins, by comparison with the fx-team branch (~2% of checkins) or mozilla-central (~4% of checkins).

mozilla-aurora, mozilla-beta:

  • ~3% of our total monthly checkins landed into mozilla-aurora, consistent with previous months.
  • ~1% of our total monthly checkins landed into mozilla-beta, consistent with previous months.

(Standard disclaimer: I’m always glad whenever we catch a problem *before* we ship a release; it avoids us having to do a chemspill release and also we ship better code to our Firefox users in the first place.)

misc other details:

  • Pushes per day

  • Pushes by hour of day

4 thoughts on “Infrastructure load for April 2012

  1. It’s also worth noting that mozilla-central and mozilla-inbound went to approvals-only for a week in there, which might further throw off stats. However, as that week was in the middle of the month, it seems likely the reduced volume for that week (17th-24th) would mostly complement increased volume for the several days after, so it might not have affected overall numbers.

    • hi Jeff;

      Good point about the approval-only for mozilla-central, mozilla-inbound. I forgot about that.

      It hard to tell what consequence, if any, that approval-only lockdown had on monthly numbers; all 30+ other branches did not have that approval-only lockdown, so might have seen increased activity during that week… and yes, as you suggest, also the week after might have seen increased activity on inbound/central because of the ending of the lockdown.

      Another (related) factor to consider in the mix could be developers who are not working on any project branch, holding back riskier patches until after FF12.0 shipped on 24april.

      Hard to tell.


Leave a Reply