Open Bug 1401172 Opened 2 years ago Updated 2 years ago
[meta] Collect telemetry data about Marionette usage
I have finally convinced a number of people(read Google) that we collecting this data would be good for the webdriver project. This bug is to track all the different parts of collecting that meta data.
Why do we have to convince Google to collect Telemetry from Firefox?
(In reply to Andreas Tolfsen ‹:ato› from comment #1) > Why do we have to convince Google to collect Telemetry from Firefox? we dont anymore but they were not going to run Selenium in their infra with telemetry pings. They, and us, turn them off for other things but getting meaningful data for webdriver is important.
Not making outside network requests during automation is a reasonable request. For certain test jobs (unfortunately not for Mn and Mn-h) we do the same by crash-asserting Firefox when it attempts a non-loopback network connection. The last time I spoke with our Telemetry people they talked about collecting certain data points in a “bucket” separate from that of non-Marionette-enabled Firefoxen. I can’t quite remember the details, but I was assured there exists ways in which we can safely enable Telemetry but not have that data count towards regular, non-instrumented, user instances of Firefox. I feel this is the point where I’ve exhausted my Telemetry knowledge and that a needinfo for someone on the data team would be prudent. As I understand it, the effect we want to achieve here is to enable Telemetry in such a way that we can collect data from Marionette, but because this data is tainted by instrumentation and automation, we don’t want it to be counted as a real data from real users.
Never fear... I have all of this in hand :D
Summary: [META] Collect telemetry data about Marionette usage → [meta] Collect telemetry data about Marionette usage
You need to log in before you can comment on or make changes to this bug.