Closed Bug 1443137 Opened 6 years ago Closed 6 years ago

Intermittent docshell/test/browser/browser_search_notification.js | Test timed out -

Categories

(Firefox :: Search, defect, P5)

Product:
Firefox
Component:
Search
Type:
defect

Tracking

()

Status:
RESOLVED FIXED
Milestone:
Firefox 64
Tracking Flags:
Tracking Status
firefox-esr60 --- fixed
firefox63 --- fixed
firefox64 --- fixed

People

(Reporter: intermittent-bug-filer, Assigned: standard8)

References

Details

(Keywords: intermittent-failure, Whiteboard: [stockwell unknown])

Attachments

(2 files)

Disable on Linux and OS X
715 bytes, patch
jmaher
: review+
Details | Diff | Splinter Review
Bug 1443137 - Attempt to fix intermittent in browser_search_notification.js by relying on a different method for opening the tab and waiting for completion before we start a search.
46 bytes, text/x-phabricator-request
adw
: review+
Details | Review 
Filed by: archaeopteryx [at] coole-files.de

https://treeherder.mozilla.org/logviewer.html#?job_id=165961591&repo=try

https://queue.taskcluster.net/v1/task/azHRZsWtTQGmRMYynFBYGg/runs/0/artifacts/public/test_info//browser-chrome-chunked_errorsummary.log

This was a central-as-beta simulation, that's a valid bug.

[task 2018-03-05T12:03:48.158Z] 12:03:48     INFO - TEST-START | docshell/test/browser/browser_search_notification.js
[task 2018-03-05T12:04:33.254Z] 12:04:33     INFO - TEST-INFO | started process screentopng
[task 2018-03-05T12:04:33.758Z] 12:04:33     INFO - TEST-INFO | screentopng: exit 0
[task 2018-03-05T12:04:33.758Z] 12:04:33     INFO - Buffered messages logged at 12:03:48
[task 2018-03-05T12:04:33.759Z] 12:04:33     INFO - TEST-PASS | docshell/test/browser/browser_search_notification.js | Check fake search engine is selected - 
[task 2018-03-05T12:04:33.760Z] 12:04:33     INFO - Console message: Security Error: Content at about:newtab may not load or link to chrome://mozapps/skin/places/defaultFavicon.svg.
[task 2018-03-05T12:04:33.760Z] 12:04:33     INFO - Console message: Security Error: Content at about:newtab may not load or link to chrome://mozapps/skin/places/defaultFavicon.svg.
[task 2018-03-05T12:04:33.763Z] 12:04:33     INFO - Console message: Security Error: Content at about:newtab may not load or link to chrome://mozapps/skin/places/defaultFavicon.svg.
[task 2018-03-05T12:04:33.764Z] 12:04:33     INFO - Buffered messages finished
[task 2018-03-05T12:04:33.765Z] 12:04:33     INFO - TEST-UNEXPECTED-FAIL | docshell/test/browser/browser_search_notification.js | Test timed out -
This only hits beta and central-as-beta simulation on Linux and OS X, except one case only opt. First failure on 2018-03-05 in a central-as-beta simulation (still gecko 60), first failure on beta 2018-03-12.

arai, could this be a regression from bug 1435908? Other suspects are bug 888600 and bug 1441964 which also landed shortly before the first failure.
Flags: needinfo?(arai.unmht)
I don't think it's from bug 1435908.

https://hg.mozilla.org/mozilla-central/rev/5c38100787fc touches handleCommand and _loadURL methods of URLBar,
that is called by the testcase, but the change should have no effect if the method is called without any parameter.
Flags: needinfo?(arai.unmht)

        Attachment #9004154 -
        Flags: review?(jmaher)

    
    

    
  
Why is this test being disabled?

Over the last two week's reports:

* try: 108
* mozilla-esr60: 1
* mozilla-release: 1
* mozilla-beta: 23

A quick glance at the failure reports for last week point me to https://treeherder.mozilla.org/#/jobs?repo=try&revision=4daf36bffc3038fc86b2c74ac2277bc35d153357&selectedJob=195969310 and https://treeherder.mozilla.org/#/jobs?repo=try&revision=cff7569e3bea876c3a2a332315163732ab50369d&selectedJob=195968587 which are both beta simulations, I suspect there are more as well.

Previous weeks have had beta at a manageable level.

I'm really not sure we should include try stats here, especially when we hit things like multiple beta simulations.
Flags: needinfo?(jmaher)
Flags: needinfo?(csabou)

    
    

    
  
Hi Mark, 
I've made a patch in case this needs to be disabled. Yes, most of the failures are on beta-sims that we run on a daily basis but also saw there are failures on mozilla-beta and that's why I decided to make a patch. I'm perfectly fine with acting against disabling this but since it has reached the disable-recommended tag maybe it should be worked on the root cause.
Flags: needinfo?(csabou)

    
    

    
  
Comment on attachment 9004154 [details] [diff] [review]
Disable on Linux and OS X

Review of attachment 9004154 [details] [diff] [review]:
-----------------------------------------------------------------

agreed this is not needed as majority of failures are on try runs.

        Attachment #9004154 -
        Flags: review?(jmaher) → review+
Flags: needinfo?(jmaher)

    
    

    
      
  

    
    

    
  
The try classifications from above are incorrect, please disregard that.

    
    

    
      
  

    
    

    
      
  

    
    

    
  
Majority of failures here occurred on try (misclassifications due to bug 1484053) and on mozilla-beta. 
Orange factor will not load a summary.

    
    

    
      
  

    
    

    
  
we have a lot of failures on mozilla-beta, we should land this patch on mozilla-beta.
Flags: needinfo?(ebalazs)

    
    

    
      
  

    
    

    
  
as this is not on trunk, I am marking as [stockwell unknown]
Whiteboard: [stockwell disable-recommended] → [stockwell unknown]

    
    

    
      
  
Whiteboard: [stockwell disable-recommended] → [stockwell unknown]

    
    

    
      
  
Whiteboard: [stockwell disable-recommended] → [stockwell unknown]

    
    

    
      
  
Whiteboard: [stockwell disable-recommended] → [stockwell unknown]

    
    

    
  
I took another look at this today, I think I have a potential fix.

My best guess is that doing a keyword search straight after addTab, is probably not necessarily a good idea - a few things might still be setting up, as addTab doesn't necessarily hand asynchronous activities. Therefore, we should use a different method of opening the browser tab.

However, what that doesn't explain is the fact this only happens when we hit beta. Still, I think this is probably an improvement in the test regardless.

I've also updated the test to make it a bit more modern, using async/await etc.

There's a try push here which has the beta simulation, plus lots of rebuilds and test verifies, and it is all green (apart from test failures not related to this one):

https://treeherder.mozilla.org/#/jobs?repo=try&revision=089240b966773bf119a9807bff6c6c43a36d9084
Assignee: nobody → standard8
Status: NEW → ASSIGNED

    
    

    
  
Comment on attachment 9011390 [details]
Bug 1443137 - Attempt to fix intermittent in browser_search_notification.js by relying on a different method for opening the tab and waiting for completion before we start a search.

Drew Willcoxon :adw has approved the revision.

        Attachment #9011390 -
        Flags: review+

    
    

    
  
Pushed by mbanner@mozilla.com:
https://hg.mozilla.org/integration/autoland/rev/f89af2a368fe
Attempt to fix intermittent in browser_search_notification.js by relying on a different method for opening the tab and waiting for completion before we start a search. r=adw

    
    

    
  
https://hg.mozilla.org/mozilla-central/rev/f89af2a368fe
Status: ASSIGNED → RESOLVED
Closed: 6 years ago

          status-firefox64:
          --- → fixed
Resolution: --- → FIXED
Target Milestone: --- → Firefox 64

    
    

    
      
  

    
    

    
      
  

          You need to log in
          before you can comment on or make changes to this bug.
        






  

    
  







  

    

      
Attachment

      

      
      
      
      
    

    

      

        

          

            
General

            

              Creator: 



            

            
Created: 

            
Updated: 

            
Size: