E10s/Status/Oct24: Difference between revisions

From MozillaWiki
< E10s‎ | Status
Jump to navigation Jump to search
(Created page with "thumbnail|left<br> =E10s Update: '''October 12'''= ==Executive Summary== * Across all release channel activity that we recorded for October 3, 30.15% of...")
 
 
(10 intermediate revisions by 2 users not shown)
Line 1: Line 1:
[[File:Green-sm.jpg|thumbnail|left]]<br>
[[File:Green-sm.jpg|thumbnail|left]]<br>


=E10s Update: '''October 12'''=
=E10s Update: '''October 24'''=
==Executive Summary==
==Executive Summary==
* Across all release channel activity that we recorded for October 3, 30.15% of the release population has e10s.  
* As of October 18, across all release channel activity we recorded, e10s is activated for 25% of the total release population (across all releases).  Although some users (about 80K) have opted out of using e10s, this number is very small compared to the total number of users on test (33.5M). See the dashboard, [https://sql.telemetry.mozilla.org/dashboard/client-count-firefox-e10s-release- here]. We are continuing to focus on resolving our two top issues to be sure we continue to trend with the majority of our users staying in the test cohort:
**Why not 40-41%?
** [https://bugzilla.mozilla.org/show_bug.cgi?id=1294719 Top crash] which has lead us to [https://bugzilla.mozilla.org/show_bug.cgi?id=1304164 temporarily disable e10s for the Russian locale] (5% of our e10s test population). '''Status:''' Patch review is ongoing and is a top priority for the networking team, we are aiming to have a fix to uplift to Firefox 51. 
***An extra 5% are being disabled because they’re running a russian locale
** [https://bugzilla.mozilla.org/show_bug.cgi?id=1300411 Slowness during tab switching]. Investigation to identify a root cause is ongoing. The approach as been combination of many efforts including of elimination (e.g. patch to interrupt GC’s and force paint appears to have had no effect on spinner occurrences) [https://bugzilla.mozilla.org/show_bug.cgi?id=1310809 and improving annotations] in background hang reporting to make BHR data actionable. See [https://docs.google.com/document/d/1MS8ol0vJPQ2eNnbZTr6lQABxRBt5KtvuTAfwlMihMls/edit?usp=sharing notes] for details.
***2.5% still on control (this is a general update problem)
* In order to better evaluate the impact e10s has on 6-week retention rates, we are going to conduct a [https://bugzilla.mozilla.org/show_bug.cgi?id=1200276 funnelcake experiment]. Stay tuned about details.
***Add-ons are sometimes blocked (stay tuned for bug #)
***4-5% on null - means they did not get the e10s roll-out
***Counting entire release population - including an extra 10% who won’t qualify (orphaned on earlier releases - not 48 or above)


==Top Issues==
=={{mrisk}} Add-Ons + WebExtensions==
* '''Investigation continues for {{Bug|1300411}} - user reports excessive load spinning when switching tabs with e10s in 48.0.2 release.''' Here is what we know:
* See full Add-Ons status, [https://wiki.mozilla.org/Firefox/AddOns/Status/current#Add-ons.2Fe10s_Program_Status_Report here]
** There are 6 individual reports; engineering is actively engaged with 1-2 people. One of which that is the most reproducible seems related to an issue with {{bug| 1279086}} google docs. [SV has been able to https://bugzilla.mozilla.org/show_bug.cgi?id=1300411#c42 manually reproduce the issue].
* Flagging the plan of record to enable MPC= true add-ons for Firefox 50 as 'at risk' as of today because there are two important uplifts that are in approval limbo. See {{bug|1309351}} and {{bug|1309350}}.
** In order to better surmise the scale of the this issue, we are landing opt-out telemetry probes as of Firefox 49 {{bug|1301104}} and {{bug|1301435}}.  
* All the right people are in engaged and a decision will be made in the next couple of days.
** We are also [https://bugzilla.mozilla.org/show_bug.cgi?id=1301131 disabling e10s] for users who are experiencing the spinner upon opening a new tab. This will be included as a ridealong for the next dot release for 49 (if we decide to ship one).
* '''We are close with a fix for our [https://bugzilla.mozilla.org/show_bug.cgi?id=1294719 top issue] that [https://bugzilla.mozilla.org/show_bug.cgi?id=1304164 caused us to temporarily disable E10s for the Russian Locale] since the crash impacts a popular email site.''' Initial tests show that the fix seems to be working as expected.
 
=={{mok}} Add-Ons + WebExtensions==
* See full Add-Ons status, [http://Firefox/AddOns/Status/current#Add-ons.2Fe10s_Program_Status_Report here]
* We shipped Firefox 49 with a set of compatible Add-Ons enabled.
* We will continue A/B experiments on Beta and will maintain the same % of the population (50% of eligible).
* General roll-out plan for including extensions is [https://docs.google.com/spreadsheets/d/1OFYrrvAFg8y1mZOEbuIR1axxtV8eul_be6ad9fe8JB0/edit?usp=sharing here].


=={{mrisk}} A11y + Windows 8 Touchscreen Support==
=={{mrisk}} A11y + Windows 8 Touchscreen Support==
* A11y and touchscreen support has landed in 51. We are targeting 52 so we can:
* A11y and touchscreen support has landed in 51. We are still targeting 52 for release and we are at the cusp of being on track but I'm staying conservative.
** Benefit from the touch events work happening  APZ
* We are basically in convergence and are tracking the issues we would like to fix before rolling out, [[Electrolysis/Accessibility#Bug_Lists|here]]
** Our primary blocker {{bug|1297549}}
*[https://bugzilla.mozilla.org/show_bug.cgi?id=1272268 Touch events for APZ has landed in 52].
** We have about [[Electrolysis/Accessibility#Bug_Lists|12 issues we would like to fix before rolling out]]
** Enabling tests
** Optimization work and [https://bugzilla.mozilla.org/show_bug.cgi?id=1272268 APZ dependency] which is slated for 52
* [http://www.nvaccess.org/ NVDA] is the first client we plan on achieving compliance with. We feel that if A11y can work with NVDA, we should be at about 90% of functionality with other, similar clients.
* [http://www.nvaccess.org/ NVDA] is the first client we plan on achieving compliance with. We feel that if A11y can work with NVDA, we should be at about 90% of functionality with other, similar clients.


Line 66: Line 52:
|9-20 || 52 default (multi-lands) || 51 default || 50 Rollout (add-ons = yes + ally = no) || 49 Rollout (known good add-ons = yes, ally = no)  
|9-20 || 52 default (multi-lands) || 51 default || 50 Rollout (add-ons = yes + ally = no) || 49 Rollout (known good add-ons = yes, ally = no)  
|-
|-
|11-8 || 53 default (multi-enabled) || 52 (multi =yes) || 51 default || 50 rollout + known good add-ons
|11-8 || 53 default (multi-enabled) || 52 (multi =yes) || 51 default || 50 rollout + known good add-ons (MPC=true)
|}
|}

Latest revision as of 05:26, 25 October 2016

Green-sm.jpg


E10s Update: October 24

Executive Summary

  • As of October 18, across all release channel activity we recorded, e10s is activated for 25% of the total release population (across all releases). Although some users (about 80K) have opted out of using e10s, this number is very small compared to the total number of users on test (33.5M). See the dashboard, here. We are continuing to focus on resolving our two top issues to be sure we continue to trend with the majority of our users staying in the test cohort:
    • Top crash which has lead us to temporarily disable e10s for the Russian locale (5% of our e10s test population). Status: Patch review is ongoing and is a top priority for the networking team, we are aiming to have a fix to uplift to Firefox 51.
    • Slowness during tab switching. Investigation to identify a root cause is ongoing. The approach as been combination of many efforts including of elimination (e.g. patch to interrupt GC’s and force paint appears to have had no effect on spinner occurrences) and improving annotations in background hang reporting to make BHR data actionable. See notes for details.
  • In order to better evaluate the impact e10s has on 6-week retention rates, we are going to conduct a funnelcake experiment. Stay tuned about details.

[AT RISK] Add-Ons + WebExtensions

  • See full Add-Ons status, here
  • Flagging the plan of record to enable MPC= true add-ons for Firefox 50 as 'at risk' as of today because there are two important uplifts that are in approval limbo. See bug 1309351 and bug 1309350.
  • All the right people are in engaged and a decision will be made in the next couple of days.

[AT RISK] A11y + Windows 8 Touchscreen Support

  • A11y and touchscreen support has landed in 51. We are still targeting 52 for release and we are at the cusp of being on track but I'm staying conservative.
  • We are basically in convergence and are tracking the issues we would like to fix before rolling out, here.
  • Touch events for APZ has landed in 52.
  • NVDA is the first client we plan on achieving compliance with. We feel that if A11y can work with NVDA, we should be at about 90% of functionality with other, similar clients.

[ON TRACK] RTL Support

  • bug 1277831 E10s support for RTL versions of Firefox on Windows and Mac is on track for Firefox 50. Softvision has just signed off on RTL support for e10s in their mid-aurora sign-offs.
  • E10s support for RTL versions of Firefox on Linux has landed for 52.

[IN PROGRESS] E10s on Windows XP

  • We are making a concerted effort to support E10s on Windows XP.
  • Details can be found, here: bug 1296279

[ON TRACK] Automated tests

[IN PROGRESS] Multi Process

Bi-weekly meetings have commenced for e10s-multi, see more info:

Release Schedule

Date Trunk Aurora Beta Release
3-07 48 default 47 default 46 A/B Tests 45 off
4-18 49 default 48 default 47 A/B Tests, Possible Rollout (add-ons = ally = no) 46 off
6-07 50 default 49 default 48 Possible Rollout (add-ons + ally = no) 47 off
8-02 51 default 50 default 49 Possible Rollout (add-ons + ally = no) 48 Rollout (add-ons + ally = no)
9-20 52 default (multi-lands) 51 default 50 Rollout (add-ons = yes + ally = no) 49 Rollout (known good add-ons = yes, ally = no)
11-8 53 default (multi-enabled) 52 (multi =yes) 51 default 50 rollout + known good add-ons (MPC=true)