https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
Jon changed:
What|Removed |Added
Status|PATCH_TO_REVIEW |RESOLVED
Resolution|---
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #67 from Gerrit Notification Bot ---
Change 126873 merged by jenkins-bot:
Use core token handling
https://gerrit.wikimedia.org/r/126873
--
You are receiving this mail because:
You are the assignee for the bug.
You are on the CC l
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
Gerrit Notification Bot changed:
What|Removed |Added
Status|NEW |PATCH_TO_REVIEW
--
You are
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #66 from Gerrit Notification Bot ---
Change 126873 had a related patch set uploaded by Jdlrobson:
Use core token handling
https://gerrit.wikimedia.org/r/126873
--
You are receiving this mail because:
You are the assignee for the
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #65 from Arthur Richards ---
(In reply to Jon from comment #64)
> We deep dived this issue again and concluded we should remove getMobileToken
> in favour of user->getToken
>
> Chris says that soon anonymous editors will get a uniq
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #64 from Jon ---
We deep dived this issue again and concluded we should remove getMobileToken in
favour of user->getToken
Chris says that soon anonymous editors will get a unique token when that is
called.
In mean time there is a
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #63 from Chris McMahon ---
The test for "Keep Going" registers a new account but also triggers the
token-mismatch problem:
https://wmf.ci.cloudbees.com/job/MobileFrontend-en.m.wikipedia.beta.wmflabs.org-linux-firefox/493/testReport/
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #62 from Chris McMahon ---
For reference, history:
http://git.wikimedia.org/history/mediawiki%2Fextensions%2FMobileFrontend/f7180dc297b958a27920a88348db27104eaf10e5/includes%2FMobileContext.php;jsessionid=7rk483w9yc4w15fi1z5t8bd05
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #61 from Jon ---
There have been 8708 hits to this error in the last 30 days on our production
wikis.
--
You are receiving this mail because:
You are the assignee for the bug.
You are on the CC list for the bug.
__
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #60 from Jon ---
The problematic function seems to be MobileContext::getMobileToken
--
You are receiving this mail because:
You are the assignee for the bug.
You are on the CC list for the bug.
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #59 from Jon ---
So this problem seems to only happen when the user is anonymous...
It seems that $this->getRequest()->getSessionData( 'wsMobileToken' ) is
returning null which results in a new session getting setup on submitting th
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #58 from Chris McMahon ---
Jon updated the editor_ve.feature test today, but I cannot get a clean run of
the test on beta labs.
In every case one token seems to persist longer than it should, on the order of
15+ seconds. I was
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
Jon changed:
What|Removed |Added
Summary|Problem with getMobileToken |Problem with getMobileToken
|i
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #57 from Chris McMahon ---
I set the editor_ve.feature test to log in with a different user anyway and
generated the following errors in the second run of the test on my local host:
2014-04-02 17:52:34 deployment-apache02 enwiki:
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #56 from Chris McMahon ---
These tests manipulate the ON/OFF toggle settings before logging in, and each
browser starts with a fresh profile on every run of every scenario. Username
should be immaterial.
--
You are receiving thi
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #55 from Chris Steipp ---
Thanks Chris, that narrows down the problem significantly.
Can you rerun the tests with a different (or new) username, so we can rule out
that this is a race condition with multiple tests using the same us
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #54 from Chris McMahon ---
From the log in beta labs on eqiad just now:
-04-02 07:46:39 deployment-apache02 enwiki:
SpecialMobileOptions::submitSettingsForm(): token mismatch, received
cb9248cad0ffc2d5886f2322ac58c2f9 - expected
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #53 from Max Semenik ---
https://gerrit.wikimedia.org/r/121879 added logging of token details to
mobile.log
--
You are receiving this mail because:
You are the assignee for the bug.
You are on the CC list for the bug.
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #52 from Chris Steipp ---
Thanks Chris. That confirms that this is different from what Jeremy posted,
since his had a Location header come back from the POST request, whereas yours
showed the form with the Error saving settings. Ple
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #50 from Jon ---
chris steipp, it's however worth noting that we've been unable to replicate
this with the above steps on OS X
--
You are receiving this mail because:
You are the assignee for the bug.
You are on the CC list for th
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #49 from Chris McMahon ---
Chris Steipp:
Pretty reliable repo for me has been to run the editor_ve.feature test, and the
second Scenario fails with the error almost every time. Watch the browser as
the test runs to see the red e
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #51 from Chris McMahon ---
Created attachment 14959
--> https://bugzilla.wikimedia.org/attachment.cgi?id=14959&action=edit
WireShark capture of error conversation
This should be the entire conversation containing the POST to Mob
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #48 from Arthur Richards ---
(In reply to Jon from comment #46)
> This is now the main source of our failing test builds (the last 3 builds
> failed solely because of this issue)
>
> Thanks for all the attention being given here.
>
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #47 from Chris Steipp ---
Is anyone able to reproduce this in the office yet? If so, I'm happy to work on
this in person tomorrow. Or if someone can walk me through the steps to
reproduce, I can attempt from a couple of installs I h
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
Jon changed:
What|Removed |Added
Priority|Unprioritized |High
--- Comment #46 from Jon ---
This is now t
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #45 from Arthur Richards ---
Chris, in the meantime, can we point the saucelabs test to not use Linux,
perhaps to use FF/Chrome in OS X since we haven't been able to replicate the
issue there?
--
You are receiving this mail becaus
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #44 from Chris McMahon ---
CSteipp, Wireshark is killing my poor underspec'd laptop. I spent a couple
hours on this, but I cannot even get the test to complete or my wlan interface
to stay functional with WireShark running.
--
Yo
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #43 from Chris Steipp ---
Chris, can you do "follow TCP Stream" and screenshot the result? I'd like to
confirm that you're NOT seeing a Location header returned in the result of that
call.
--
You are receiving this mail because:
Y
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #42 from Chris McMahon ---
The screen shot above I think shows the token-mismatch problem on the POST.
However, the test itself behaves strangely on my local machine with WireShark
running so I may have done this incorrectly
--
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #41 from Chris McMahon ---
Created attachment 14943
--> https://bugzilla.wikimedia.org/attachment.cgi?id=14943&action=edit
token mismatch on POST shown in WireShark on Linux
--
You are receiving this mail because:
You are the as
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
Chris Steipp changed:
What|Removed |Added
CC||cste...@wikimedia.org
--- Comment #40 f
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
Jon changed:
What|Removed |Added
Status|PATCH_TO_REVIEW |NEW
--
You are receiving this mail because:
You
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #39 from Jeremy L Grajales ---
Created attachment 14914
--> https://bugzilla.wikimedia.org/attachment.cgi?id=14914&action=edit
(Partial) Wireshark capture of editor_ve test
[attachments]
I captured the test with Wireshark.
The PO
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #38 from Chris McMahon ---
Requested by Chris Steipp, Request headers and stuff pulled from the Network
tab in Firefox after incurring the token error (I also have the entire response
that contains the error message if you need it)
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #37 from Jeremy L Grajales ---
(In reply to Arthur Richards from comment #36)
> Chris, any luck dumping those headers? I'm increasingly curious to see
> what's going on there.
What headers are you looking for at what point?
I'd lik
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #36 from Arthur Richards ---
(In reply to Chris McMahon from comment #27)
> I chatted with Chris Steipp about this a bit:
>
> (03:53:51 PM) csteipp: So yeah, either the user doesn't have a session,
> wsMobileToken in the session i
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #35 from Gerrit Notification Bot ---
Change 120959 abandoned by Jdlrobson:
See if sleeping helps avoid token issues on Special:MobileOptions
Reason:
Thanks to Jeremy L Grajales we can confirm this doesn't have any effect. Back
to d
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #33 from Jeremy L Grajales ---
(In reply to Gerrit Notification Bot from comment #32)
> Change 120959 had a related patch set uploaded by Jdlrobson:
> See if sleeping helps avoid token issues on Special:MobileOptions
>
> https://ge
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #34 from Jeremy L Grajales ---
(In reply to Gerrit Notification Bot from comment #32)
> Change 120959 had a related patch set uploaded by Jdlrobson:
> See if sleeping helps avoid token issues on Special:MobileOptions
>
> https://ge
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
Gerrit Notification Bot changed:
What|Removed |Added
Status|UNCONFIRMED |PATCH_TO_REVIEW
Ever con
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #32 from Gerrit Notification Bot ---
Change 120959 had a related patch set uploaded by Jdlrobson:
See if sleeping helps avoid token issues on Special:MobileOptions
https://gerrit.wikimedia.org/r/120959
--
You are receiving this m
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #31 from Jeremy L Grajales ---
(In reply to Jon from comment #30)
> If the opt into alpha happens after login does this resolve the issue? I
> wonder if it's something particularly strange with opting in whilst
> anonymous...
Don't
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #30 from Jon ---
If the opt into alpha happens after login does this resolve the issue? I wonder
if it's something particularly strange with opting in whilst anonymous...
--
You are receiving this mail because:
You are the assigne
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #29 from Jeremy L Grajales ---
(In reply to Jon from comment #9)
> If someone can help me get Chrome driver setup maybe I can find a way around
> this... does delaying the opt in event with a wait help at all Chris?
I also added a
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #28 from Jeremy L Grajales ---
I tested this on Firefox (Windows).
http://pastebin.com/yi9FGPVU
Seems to reproduce the same issue.
Although, I'm not sure.
Visually:
The browser gets to the settings page.
Toggles beta switch ON.
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #27 from Chris McMahon ---
I chatted with Chris Steipp about this a bit:
(03:53:51 PM) csteipp: So yeah, either the user doesn't have a session,
wsMobileToken in the session is getting overwritten, or the form isn't
correctly sen
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #26 from Arthur Richards ---
(In reply to Arthur Richards from comment #23)
> I agree with Max, in this case a race condition doesn't make sense due to
> the nature of PHP. However, it is conceivable that there are occasional
> hicc
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #25 from Arthur Richards ---
Jeremy if you (In reply to Jeremy L Grajales from comment #24)
> Is there anything I could do to attempt to replicate this on my end?
Thanks Jeremy - if you don't have it set up already, the first thing
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
Jeremy L Grajales changed:
What|Removed |Added
CC||jeremygraja...@gmail.com
--- Comme
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #20 from Max Semenik ---
Chris, these are synchronous operations, it's really hard to do stufff in
parallel in PHP:) So no race condition.
--
You are receiving this mail because:
You are the assignee for the bug.
You are on the CC
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #22 from Chris McMahon ---
I think the fact that this is passing anywhere is an environment issue.
Same error happened twice in Firefox in the build just hours ago:
https://wmf.ci.cloudbees.com/job/MobileFrontend-en.m.wikipedia.
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #21 from Jon ---
As stated before, I really thing this is an environment issue. Chris what else
might be different in your setup to us?
--
You are receiving this mail because:
You are the assignee for the bug.
You are on the CC li
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #23 from Arthur Richards ---
I agree with Max, in this case a race condition doesn't make sense due to the
nature of PHP. However, it is conceivable that there are occasional
hiccups/something going wrong with session storage on the
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #19 from Chris McMahon ---
Seems like a race condition in lines 204-208 of SpecialMobileOptions.php in
submitSettingsForm.
As I read this:
* submitSettingsForm calls out to do "$context = MobileContext::singleton();"
and also "
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #18 from Jon ---
Yep but still unable to replicate this bug in any way either locally or on beta
labs...
--
You are receiving this mail because:
You are the assignee for the bug.
You are on the CC list for the bug.
___
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #17 from Ċ½eljko Filipin ---
(In reply to Jon from comment #9)
> If someone can help me get Chrome driver setup maybe I can find a way around
> this... does delaying the opt in event with a wait help at all Chris?
Jon, did you manag
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #16 from Chris McMahon ---
and here is the same problem in Firefox, so it is not only Chrome:
https://wmf.ci.cloudbees.com/job/MobileFrontend-en.m.wikipedia.beta.wmflabs.org-linux-firefox/430/testReport/(root)/VisualEditor/Toolbar_
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #15 from Chris McMahon ---
ChromeDriver (v2.9.248304)
but that has nothing to do with the problem.
--
You are receiving this mail because:
You are the assignee for the bug.
You are on the CC list for the bug.
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #14 from Jon ---
I understand - I can see this in the browser test screencast but I'm not seeing
this so your environment must be setup slightly different to mine?
What version of Chrome Driver are you using?
--
You are receiving
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #13 from Chris McMahon ---
Created attachment 14852
--> https://bugzilla.wikimedia.org/attachment.cgi?id=14852&action=edit
error saving settings
--
You are receiving this mail because:
You are the assignee for the bug.
You are o
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #12 from Jon ---
I understand but I ran exactly the same test 6 times and can't replicate this.
This always passes for me.
What version of Chrome Driver are you using?
--
You are receiving this mail because:
You are the assignee
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #11 from Chris McMahon ---
I got the problem the first time I ran, on the second scenario:
$ bundle exec cucumber features/editor_ve.feature
@en.m.wikipedia.beta.wmflabs.org @login @test2.m.wikipedia.org
Feature: VisualEditor
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #10 from Jon ---
I have run this 6 times locally against beta labs with the latest ChromeDriver
(v2.9.248307) and all passed.
I'm still not sure what's going on here.
--
You are receiving this mail because:
You are the assignee f
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #9 from Jon ---
If someone can help me get Chrome driver setup maybe I can find a way around
this... does delaying the opt in event with a wait help at all Chris?
--
You are receiving this mail because:
You are the assignee for th
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #8 from Chris McMahon ---
To the extent that it is an issue with Selenium, the speed of the test between
clicking the "ON" toggle and the "Save" button is probably triggering a race
condition that is difficult to trigger manually.
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #7 from Arthur Richards ---
I wonder if it's possibly an issue with Selenium? This is not something I've
been able to repro either and certainly not something I've experienced locally
or in production.
--
You are receiving this ma
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #6 from Chris McMahon ---
Max, for me when I run the browser test "editor_ve.feature" (or any test with
the step "Given I am in alpha mode") against beta labs, I see the "Error
saving settings. Please make sure that you have cookie
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
--- Comment #5 from Max Semenik ---
Chris, how do I repro this problem locally?
--
You are receiving this mail because:
You are the assignee for the bug.
You are on the CC list for the bug.
___
Wiki
https://bugzilla.wikimedia.org/show_bug.cgi?id=62614
Jon changed:
What|Removed |Added
Summary|intermittent: attempt to|Problem with getMobileToken
|t
69 matches
Mail list logo
