In about 15% of pings coming from recent Nightly builds the e10sEnabled flag doesn't match the E10S_WINDOW histogram [1]. [1] https://gist.github.com/vitillo/6e7a04f8b92b2bcc2ffd

In bug 1179189, we switched over e10sEnabled to using Services.appinfo.browserTabsRemoteAutostart. E10S_WINDOW submits (AFAICT) whether e10s is enabled per-window: https://dxr.mozilla.org/mozilla-central/rev/9ed17db42e3e46f1c712e4dffd62d54e915e0fac/browser/base/content/browser.js#1279 https://dxr.mozilla.org/mozilla-central/rev/9ed17db42e3e46f1c712e4dffd62d54e915e0fac/browser/base/content/browser.js#94 If we still have any features to open e10s enabled/disabled windows then that mismatch is expected. mconley, does that sound right?

I think Roberto compared e10sEnabled vs E10S_AUTOSTART instead, no?

(In reply to Vladan Djeric (:vladan) -- please needinfo! from comment #2) > I think Roberto compared e10sEnabled vs E10S_AUTOSTART instead, no? Yeah I did, I actually started with E10S_WINDOW and later switched to E10S_AUTOSTART. Either way though, the conclusion is nearly the same.

Yes, gfritzsche is right - the mismatch between e10sEnabled and E10S_WINDOW is expected. E10S_WINDOW is incremented every time an e10s-enabled window is opened, and it is still very possible to open non-e10s windows. The mismatch between e10sEnabled and E10S_AUTOSTART is less expected. It actually looks like they're measuring the same things. I assume that TelemetryEnvironment is only reading this value once - and that's what's occurring with BrowserTabsRemoteAutostart: https://dxr.mozilla.org/mozilla-central/rev/9ed17db42e3e46f1c712e4dffd62d54e915e0fac/toolkit/xre/nsAppRunner.cpp#4611 (notice that it's memoized) I agree with your results - there is a mismatch. I cannot explain where that mismatch is coming from. :( Are there pings where e10sEnabled is true, but E10S_AUTOSTART is not? Or the other way around? Or both?

(In reply to Mike Conley (:mconley) - Needinfo me! from comment #4) > I agree with your results - there is a mismatch. I cannot explain where that > mismatch is coming from. :( > > Are there pings where e10sEnabled is true, but E10S_AUTOSTART is not? Or the > other way around? Or both? It appears that the mismatch is caused exclusively by pings for which e10sEnabled is true and E10S_AUTOSTART is false.

(In reply to Roberto Agostino Vitillo (:rvitillo) from comment #5) > (In reply to Mike Conley (:mconley) - Needinfo me! from comment #4) > > I agree with your results - there is a mismatch. I cannot explain where that > > mismatch is coming from. :( > > > > Are there pings where e10sEnabled is true, but E10S_AUTOSTART is not? Or the > > other way around? Or both? > > It appears that the mismatch is caused exclusively by pings for which > e10sEnabled is true and E10S_AUTOSTART is false. Interesting... and for those cases, is E10S_AUTOSTART_STATUS set to a consistent value?

E10S_AUTOSTART_STATUS is always empty for those cases.

(In reply to Roberto Agostino Vitillo (:rvitillo) from comment #7) > E10S_AUTOSTART_STATUS is always empty for those cases. That's very puzzling. Are there any other patterns you notice from those pings? A shared OS, for example?

The only pattern I could notice is that the median subsession duration is about 5 times longer compared to the general population, but I am afraid it isn't very useful in tracking down the cause.

Is this something we still need to fix or do we have this data in other places?

Afaik this has not been fixed.

E10S_AUTOSTART was removed in bug 1111701. E10S_WINDOW has never been in use. Lets convert this into a bug about removing the window probe. http://mxr.mozilla.org/mozilla-central/source/toolkit/components/telemetry/Histograms.json#8512

This bug is about removing the histogram "E10S_WINDOW" and related code: https://dxr.mozilla.org/mozilla-central/search?q=E10S_WINDOW&case=true

I think no one is looking at this probe. This measurement would be interesting to look at for users who don't have it to 100% in one or the other bucket (i.e., are there things causing non-e10s windows to be opened by accident). But I don't know if that's a real concern as I've never seen a bug reported on that.

Vedant, would you be interested in taking this one?

(In reply to Alessio Placitelli [:Dexter] from comment #15) > Vedant, would you be interested in taking this one? Sure Sir. I'll start work on this ASAP.

I also ran |./mach test toolkit/components/telemetry/|. All tests were successful.

Comment on attachment 8828905 [details] [diff] [review] Proposed patch for bug 1206048. Review of attachment 8828905 [details] [diff] [review]: ----------------------------------------------------------------- Thanks for the patch, this looks good Vedant, good job! Let me push this to try for you before checking it in.

Looks good in try, let's land this!

Pushed by ryanvm@gmail.com: https://hg.mozilla.org/integration/mozilla-inbound/rev/3aa2f67fa2bc Remove E10S_WINDOW metric. r=Dexter

(In reply to Alessio Placitelli [:Dexter] from comment #18) > ----------------------------------------------------------------- > > Thanks for the patch, this looks good Vedant, good job! > > Let me push this to try for you before checking it in. Thank you! :)