1388160 - Disable predictor prefetch (again!)

Status: RESOLVED FIXED

(Core :: Networking, defect)

Description

[u408661]

We need to do this on nightly 57 and beta 56, based on a new email from Y! saying the previous fix didn't entirely fix the issue on their end.

Comment 1

[u408661]

Attachment: Bug 1388160 - Disable prefetch (again).

Review commit: https://reviewboard.mozilla.org/r/165854/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/165854/

Comment 2

[Jason Duell]

Comment on attachment 8894692 [details]
Bug 1388160 - Disable prefetch (again).

https://reviewboard.mozilla.org/r/165854/#review171022

Comment 3

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/ce2b6a193210
Disable prefetch (again). r=jduell

Comment 4

[u408661]

Comment on attachment 8894692 [details]
Bug 1388160 - Disable prefetch (again).

Approval Request Comment
[Feature/Bug causing the regression]: predictor prefetch
[User impact if declined]: causes unwanted duplicate requests to ad networks
[Is this code covered by automated tests?]: yes
[Has the fix been verified in Nightly?]: yes (just not this particular landing of the pref flip)
[Needs manual test from QE? If yes, steps to reproduce]: nope
[List of other uplifts needed for the feature/fix]: none
[Is the change risky?]: no
[Why is the change risky/not risky?]: simple pref flip, that's been flipped before
[String changes made/needed]: none

Comment 5

[u408661]

Comment on attachment 8894692 [details]
Bug 1388160 - Disable prefetch (again).

See above for approval request comment. Requesting approval for release so this can hit any dot release we make for 55. We have bug 1388158 to send out a pref flip add-on to handle 55.0, so we shouldn't have to make a dot release just for this :)

Comment 6

[Ionuț Goldan [:igoldan]]

Improved noticed due to this change:

== Change summary for alert #8637 (as of August 08 2017 17:04 UTC) ==

Improvements:

  2%  tp5o_webext summary windows7-32 opt e10s     331.84 -> 324.03

For up to date results, see: https://treeherder.mozilla.org/perf.html#/alerts?id=8637

Comment 7

[Sylvestre Ledru [:Sylvestre]]

Comment on attachment 8894692 [details]
Bug 1388160 - Disable prefetch (again).

OK, let's take it for 55.0.1 and for 56.0beta2.

Comment 8

[Sylvestre Ledru [:Sylvestre]]

Nicholas or Jason, for posterity, could you give us the bug number in which we disabled it the first time? thanks
(I tried to needinfo Nicholas but he is refusing them)

Comment 9

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/ce2b6a193210

Comment 10

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/releases/mozilla-beta/rev/c777f07ebf3ed32ca2d893b155ff6dca4b0638b1 for beta
 https://hg.mozilla.org/releases/mozilla-release/rev/e56fed89e8ac27815207bfbbaccb0ea90a57de7e for release

Comment 11

[Sylvestre Ledru [:Sylvestre]]

was it available on firefox for android?
Thanks

Comment 12

[Sylvestre Ledru [:Sylvestre]]

Dragana answered to my question:
<dragana> Sylvestre: no, it is not desktop only. I do not see any #ifdefs and there is no such a pref entry in the android only prefs file

Comment 13

[Jason Duell]

We've disabled it at least a few times.  Bugzilla is being very weird for me (search for "predictor" doesn't turn up many bugs, but "Disable predictor prefetch" turns up bug 1268208 and bug 1278636.

Comment 14

[Alexandru Simonca [:asimonca], Release Desktop QA]

the pref "network.predictor.enable-prefetch" is set to "false" in 55.0.1 build2 and Nightly 57.0a1 on the following OSs:
- Windows 10 x64
- Windows 7 x64
- Ubuntu 16.04 x86
- macOS 10.12.6

Comment 15

[Sylvestre Ledru [:Sylvestre]]

We will build a 55.0.1 for fennec, so, it will be disabled in this product too.