WebKit Bugzilla
New
Browse
Log In
×
Sign in with GitHub
or
Remember my login
Create Account
·
Forgot Password
Forgotten password account recovery
RESOLVED FIXED
154595
[GTK] Run the run-benchmark script on the performance bot.
https://bugs.webkit.org/show_bug.cgi?id=154595
Summary
[GTK] Run the run-benchmark script on the performance bot.
Carlos Alberto Lopez Perez
Reported
2016-02-23 10:57:47 PST
After
bug 153993
is possible to use the run-benchmark script to execute the benchmarks with the WebKitGTK+ MiniBrowser Now we would to implement a new step for the performance bots to run this tests automatically (running all plans available). The idea is to only enable this for the GTK+ Performance Test bot for the moment. Other ports can follow if they wish. Previous context:
https://lists.webkit.org/pipermail/webkit-dev/2016-February/027948.html
Attachments
Patch
(12.27 KB, patch)
2016-03-01 08:35 PST
,
Carlos Alberto Lopez Perez
no flags
Details
Formatted Diff
Diff
View All
Add attachment
proposed patch, testcase, etc.
Carlos Alberto Lopez Perez
Comment 1
2016-03-01 08:35:17 PST
Created
attachment 272564
[details]
Patch
WebKit Commit Bot
Comment 2
2016-03-01 08:37:07 PST
Attachment 272564
[details]
did not pass style-queue: ERROR: Tools/BuildSlaveSupport/build.webkit.org-config/mastercfg_unittest.py:534: [RunBenchmarkTest.assertResults] Undefined variable 'RunBenchmarkTests' [pylint/E0602] [5] Total errors found: 1 in 6 files If any of these errors are false positives, please file a bug against check-webkit-style.
Carlos Alberto Lopez Perez
Comment 3
2016-03-01 08:39:39 PST
(In reply to
comment #2
)
>
Attachment 272564
[details]
did not pass style-queue: > > > ERROR: > Tools/BuildSlaveSupport/build.webkit.org-config/mastercfg_unittest.py:534: > [RunBenchmarkTest.assertResults] Undefined variable 'RunBenchmarkTests' > [pylint/E0602] [5] > Total errors found: 1 in 6 files > > > If any of these errors are false positives, please file a bug against > check-webkit-style.
This is a false positive. The unit tests are executed as follows: ( cd Tools/BuildSlaveSupport/build.webkit.org-config/; ./mastercfg_unittest.py )
Carlos Garcia Campos
Comment 4
2016-03-03 08:10:36 PST
Comment on
attachment 272564
[details]
Patch Looks good, let's try it out! Please, watch the bots after landing this.
Csaba Osztrogonác
Comment 5
2016-03-03 08:20:17 PST
(In reply to
comment #4
)
> Comment on
attachment 272564
[details]
> Patch > > Looks good, let's try it out! Please, watch the bots after landing this.
You will need a buildmaster restart to see its effect.
Carlos Alberto Lopez Perez
Comment 6
2016-03-03 08:40:15 PST
Comment on
attachment 272564
[details]
Patch Clearing flags on attachment: 272564 Committed
r197505
: <
http://trac.webkit.org/changeset/197505
>
Carlos Alberto Lopez Perez
Comment 7
2016-03-03 08:40:22 PST
All reviewed patches have been landed. Closing bug.
Carlos Alberto Lopez Perez
Comment 8
2016-03-03 10:26:46 PST
(In reply to
comment #5
)
> (In reply to
comment #4
) > > Comment on
attachment 272564
[details]
> > Patch > > > > Looks good, let's try it out! Please, watch the bots after landing this. > > You will need a buildmaster restart to see its effect.
yes, it seems. I have sent a mail to lucas asking for a restart.
Carlos Alberto Lopez Perez
Comment 9
2016-03-03 10:45:27 PST
Lucas just restarted the master moments ago :) I can see now the new step on the GTK+ perf bot. Let's see how it goes
Carlos Alberto Lopez Perez
Comment 10
2016-03-03 15:57:11 PST
Committed
r197525
: <
http://trac.webkit.org/changeset/197525
>
Carlos Alberto Lopez Perez
Comment 11
2016-03-05 11:44:36 PST
Committed
r197615
: <
http://trac.webkit.org/changeset/197615
>
Csaba Osztrogonác
Comment 12
2017-07-31 14:46:33 PDT
Comment on
attachment 272564
[details]
Patch View in context:
https://bugs.webkit.org/attachment.cgi?id=272564&action=review
> Tools/BuildSlaveSupport/build.webkit.org-config/master.cfg:712 > +class RunBenchmarkTests(shell.Test): > + name = "benchmark-test" > + description = ["benchmark tests running"] > + descriptionDone = ["benchmark tests"] > + # Buildbot default timeout without output for a step is 1200. > + # The current maximum timeout for a benchmark plan is also 1200. > + # So raise the buildbot timeout to avoid aborting this whole step when a test timeouts. > + timeout = 1500 > + command = ["python", "./Tools/Scripts/run-benchmark", "--allplans"]
Setting timeout here doesn't have any effect, see the buildbot logs:
https://build.webkit.org/builders/GTK%20Linux%2064-bit%20Release%20%28Perf%29/builds/831/steps/benchmark-test/logs/stdio
The timeout is still the default 1200 seconds. You need a fix like this:
https://trac.webkit.org/changeset/179043/webkit
Note
You need to
log in
before you can comment on or make changes to this bug.
Top of Page
Format For Printing
XML
Clone This Bug