697555 - when turning on responsiveness for tp5, the browser hangs or crashes

Status: RESOLVED FIXED

(Testing :: Talos, defect)

Description

[Joel Maher ( :jmaher ) (UTC -8) (PTO back normal Nov 17)]

On osx 10.5.8 and 10.6.2 we fail to terminate the browser after completing the tp5 test.  Similarly on Fedora 12, we end up crashing.


linux:
https://tbpl.mozilla.org/php/getParsedLog.php?id=7041789&tree=Mozilla-Inbound

10.6.2:
https://tbpl.mozilla.org/php/getParsedLog.php?id=7040940&tree=Mozilla-Beta

10.5.8:
https://tbpl.mozilla.org/php/getParsedLog.php?id=7041676&tree=Mozilla-Aurora

Comment 1

[Joel Maher ( :jmaher ) (UTC -8) (PTO back normal Nov 17)]

so it seems as though similar symptoms exist without this, just much more infrequently:
https://bugzilla.mozilla.org/show_bug.cgi?id=664371

Comment 2

[Joel Maher ( :jmaher ) (UTC -8) (PTO back normal Nov 17)]

ok, I just realized that the OSX failures were on aurora/beta which do not have the fix that ted landed.

Comment 3

[Joel Maher ( :jmaher ) (UTC -8) (PTO back normal Nov 17)]

Attachment: allow for responsiveness to be controlled by a commandline flag (1.0)

this will allow us to control responsiveness via the commandline from buildbot.

Comment 4

[Jeff Hammel]

Comment on attachment 572115 [details] [diff] [review]
allow for responsiveness to be controlled by a commandline flag (1.0)

+            if self.responsiveness and ('responsiveness' in line):
+                newline = line.replace('False', 'True')
+

I don't really like this as it tightly couples the format of the file with the PerfConfigurator.  Old news, I know.  That said, this should all be cleaned up so I think its fine for now.

Comment 5

[(not currently active) Ted Mielczarek]

I think this will fix the OS X issues, but we'll still have the GTK issue, right?

Comment 6

[Joel Maher ( :jmaher ) (UTC -8) (PTO back normal Nov 17)]

I haven't been able to reproduce the GTK issue.  That could have been a one off error on linux/firefox/tp5 unrelated to this.

Comment 7

[Joel Maher ( :jmaher ) (UTC -8) (PTO back normal Nov 17)]

landed the talos change:
http://hg.mozilla.org/build/talos/rev/e714827eece1

Comment 8

[Joel Maher ( :jmaher ) (UTC -8) (PTO back normal Nov 17)]

Attachment: disable responsiveness for linux (1.0)

since we are seeing some browsers exiting with non zero return codes in our staging environment, lets not run tests on linux until we can figure that out.

Comment 9

[William Lachance (:wlach)]

Comment on attachment 575186 [details] [diff] [review]
disable responsiveness for linux (1.0)

+            if 'responsiveness' in test_config and test_config['responsiveness'] and platform.system() != "Linux":

Couldn't you just write this as:

test_config.get('responsiveness') and platform.system() != "Linux":

(same for the one below)

Otherwise, looks fine.

Comment 10

[Joel Maher ( :jmaher ) (UTC -8) (PTO back normal Nov 17)]

pushed to talos:
http://hg.mozilla.org/build/talos/rev/9baf50c14041

Comment 11

[Joel Maher ( :jmaher ) (UTC -8) (PTO back normal Nov 17)]

this has landed and has had a few data points successfully

Comment 12

[Jeff Hammel]

So I notice this comment at http://hg.mozilla.org/build/talos/file/705a12151bf2/talos/ttest.py#l256 :

# ignore responsiveness tests on linux until we fix Bug 697555

Since this bug is now resolved->fixed , should we remove this code block?  Should we reopen this bug for this purpose or file another one?

Comment 13

[cmtalbert]

(In reply to Jeff Hammel [:jhammel] from comment #12)
> So I notice this comment at
> http://hg.mozilla.org/build/talos/file/705a12151bf2/talos/ttest.py#l256 :
> 
> # ignore responsiveness tests on linux until we fix Bug 697555
> 
> Since this bug is now resolved->fixed , should we remove this code block? 
> Should we reopen this bug for this purpose or file another one?

File a new one to turn on responsiveness for linux please.

Comment 14

[Jeff Hammel]

done: bug 710296