Bug 306580 – REGRESSION(306358@main)[GTK]: Caused 2 new API test timeouts: WebKitGTK/TestWebKitFaviconDatabase:/webkit/WebKitFaviconDatabase/ephemeral and WebKitGTK/TestWebKitFaviconDatabase:/webkit/WebKitFaviconDatabase/clear

• Component: WebKitGTK
• Status: RESOLVED DUPLICATE

[Carlos Alberto Lopez Perez]

Since 306358@main the GTK API tests are failing unexpectedly with:

Unexpected timeouts (2)
    /WebKitGTK/TestWebKitFaviconDatabase
        /webkit/WebKitFaviconDatabase/ephemeral
        /webkit/WebKitFaviconDatabase/clear


This was warned clearly on the EWS run at https://github.com/WebKit/WebKit/pull/57399

See: 

 - Last good run: https://build.webkit.org/#/builders/57/builds/23022
 - First bad run  : https://build.webkit.org/#/builders/57/builds/23023

[Carlos Alberto Lopez Perez]

BTW, I double checked this locally.


1. At 306358@main 
$ Tools/Scripts/run-gtk-tests --release --json-output=api_test_results.json TestWTF:WTF.DragonBox WebKitGTK/TestWebKitFaviconDatabase:/webkit/WebKitFaviconDatabase/clear WebKitGTK/TestWebKitFaviconDatabase:/webkit/WebKitFaviconDatabase/ephemeral
[...]
Unexpected timeouts (2)
    /WebKitGTK/TestWebKitFaviconDatabase
        /webkit/WebKitFaviconDatabase/clear
        /webkit/WebKitFaviconDatabase/ephemeral


2. At 306357@main 
$ Tools/Scripts/run-gtk-tests --release --json-output=api_test_results.json TestWTF:WTF.DragonBox WebKitGTK/TestWebKitFaviconDatabase:/webkit/WebKitFaviconDatabase/clear WebKitGTK/TestWebKitFaviconDatabase:/webkit/WebKitFaviconDatabase/ephemeral
[...]
Ran 3 tests of 3 with 3 successful

[Adrian Perez]

(In reply to Carlos Alberto Lopez Perez from comment #0)
> Since 306358@main the GTK API tests are failing unexpectedly with:
> 
> Unexpected timeouts (2)
>     /WebKitGTK/TestWebKitFaviconDatabase
>         /webkit/WebKitFaviconDatabase/ephemeral
>         /webkit/WebKitFaviconDatabase/clear
> 
> 
> This was warned clearly on the EWS run at
> https://github.com/WebKit/WebKit/pull/57399

The patch only changes the logging statements, and therefore it involves no change in behaviour.

> See: 
> 
>  - Last good run: https://build.webkit.org/#/builders/57/builds/23022
>  - First bad run  : https://build.webkit.org/#/builders/57/builds/23023

My suspicion is that the tests cases have been flaky, while working on a new feature I have found a number of issues with the code, fixed e.g. by landing https://github.com/WebKit/WebKit/pull/57397 (where tests passed), preparing a new fix in https://github.com/WebKit/WebKit/pull/57515 (also passed), and currently I am trying to figure out why the “get-favicon” test case times out (it's skipped in the test expectations).

[Carlos Alberto Lopez Perez]

BTW, I double checked this locally.


1. At 306358@main 
$ Tools/Scripts/run-gtk-tests --release WebKitGTK/TestWebKitFaviconDatabase:/webkit/WebKitFaviconDatabase/clear WebKitGTK/TestWebKitFaviconDatabase:/webkit/WebKitFaviconDatabase/ephemeral
[...]
Unexpected timeouts (2)
    /WebKitGTK/TestWebKitFaviconDatabase
        /webkit/WebKitFaviconDatabase/clear
        /webkit/WebKitFaviconDatabase/ephemeral


2. At 306357@main 
$ Tools/Scripts/run-gtk-tests --release WebKitGTK/TestWebKitFaviconDatabase:/webkit/WebKitFaviconDatabase/clear WebKitGTK/TestWebKitFaviconDatabase:/webkit/WebKitFaviconDatabase/ephemeral
[...]
Ran 2 tests of 2 with 2 successful

[Carlos Alberto Lopez Perez]

(In reply to Adrian Perez from comment #2)
> 
> My suspicion is that the tests cases have been flaky, while working on a new
> feature I have found a number of issues with the code, fixed e.g. by landing
> https://github.com/WebKit/WebKit/pull/57397 (where tests passed), preparing
> a new fix in https://github.com/WebKit/WebKit/pull/57515 (also passed), and
> currently I am trying to figure out why the “get-favicon” test case times
> out (it's skipped in the test expectations).

I can 100% reproduce this locally. At 306358@main always timeout and at 306357@main always passes. See comment above about how to reproduce it

[Adrian Perez]

(In reply to Carlos Alberto Lopez Perez from comment #4)
> (In reply to Adrian Perez from comment #2)
> > 
> > My suspicion is that the tests cases have been flaky, while working on a new
> > feature I have found a number of issues with the code, fixed e.g. by landing
> > https://github.com/WebKit/WebKit/pull/57397 (where tests passed), preparing
> > a new fix in https://github.com/WebKit/WebKit/pull/57515 (also passed), and
> > currently I am trying to figure out why the “get-favicon” test case times
> > out (it's skipped in the test expectations).
> 
> I can 100% reproduce this locally. At 306358@main always timeout and at
> 306357@main always passes. See comment above about how to reproduce it

It's this missing check that was added, I had already forgotten that there was that additional change on top of the improved logging statements:

  https://github.com/WebKit/WebKit/commit/8accbbaa5baa01387b788b82fc40501662397b02#diff-e1b15fc026875945ec262f448a7dc903667dc6d356175370a0d9e380f9a67345R70

The issue is solved by https://github.com/WebKit/WebKit/pull/57515 which avoids trying to use a table from the database when it does not exist.

What I find puzzling is why the run at https://github.com/WebKit/WebKit/pull/57397 shows api-test as passed ¯\_(ツ)_/¯

[Carlos Alberto Lopez Perez]

Thanks for the fix!

Closing. This was fixed at bug 306568

*** This bug has been marked as a duplicate of bug 306568 ***