Firefox/Push/Progress Report/2017Q3

From MozillaWiki
Jump to: navigation, search

September 27, 2017

Q3 OKR Progress

Objective Key Result Confidence Features Notes
July 19 Aug 2 Aug 16 Aug 30 Sep 13 Score
1. Provide a more useful testing environment for WebPush. 1.1 Create an API Layer for Ardere. 50% 50% 50% 50% 20% 0.0

API Layer for Ardere

  • Priority went to other OKR’s as Ardere is functioning acceptably for QA and our own uses at the moment.
1.2 Create a "smoke-test" for the service, a quick test that verifies it’s up and running. 50% 50% 80% 80% 80% 0.5

Smoketest to verify Push is up and running

  • Getting a verification from QA today regarding whether or not this is addressing their needs.
  • Met with chris hartjes in the afternoon and got approval.
1.3 Github/Jenkins API call integration with Ardere 50% 50% 50% 50% 20% 0.0

Github/Jenkins API call integration with Ardere

  • Dependent upon 1.1
2. Better understanding of production run environment 2.1 Detailed heap memory consumption report as the OS sees memory under load. 60% 65% 70% 70% 80% 0.5

Detailed heap memory consumption report

  • Have detailed memory consumption. Getting the memory information we wanted.
  • Phil is preparing the report by the end of the week.
  • Loftier goals of easily reproducible load test setups (ardere, debug symbols more easily available) not quite achieved (hence 0.5)
2.2 Code in PyPy that allows us to dump detailed object graphs of allocated Python objects.

Note: Contractor is expected to help with this work. This involves a breakdown of what PyPy is doing.

50% 50% 52% 52% 60% .299

Detailed object graphs of allocated Python objects

  • Just now under progress because it took so long for contract to be finalized with the contractor.
  • Maybe a 0.1 -- but scoring higher for all the effort of moving the contract forward.
2.3 Create a graph showing the delta for memory consumption over time on a per-release basis. 80% 90% 90% 90% 90% 0.5

Memory Consumption Per Release Graph

  • Don’t have the detailed breakdown by version we had hoped for, but we do have a usable dashboard that shows us the information we want over time. (tool issue)
  • Unfortunately this will be going away when we migrate off Datadog, so work will need to be done again to make the data available in QuickSight.
3. Reduce costs 3.1 Create and implement a Rust PoC plan. 80% 80% 90% 90% 100% 1.0

Rust PoC Plan

  • Full plan was made, implemented, and even load-tested both with and without SSL.
  • It was awesome and came out as well as we’d hoped. \o/
  • And kudos to Alex Crichton.
3.2 Spec a Rust <-> Python API of Rust PoC. 80% 100% 100% 100% 100% 1.0

Rust <-> Python API of Rust PoC

  • Done!
  • Spec’d out, with preliminary implementation laid out for us by Alex Crichton.
3.3 Kill Simplepush. 60% 90% 90% 93% 100% 1.0

Retirement of Simplepush

  • Deployed to production and no issues reported.
3.4 Relax VAPID checks for partner endpoints. 60% 100% 100% 100% 100% 0.7

Relaxing of VAPID checks for partner endpoints.

  • Done!
  • Went exactly as expected.