Hi, Sahoo
> I agree Hudson was not reporting failures correctly, but that's
> because QL script was not generating proper TestNG reports. Kohsuke
> identified the problem and fixed it right away. But, what surprises me
> is that subsequent committers have not even noticed the failures in
> their environment. Let me ask you how you ensure that QL has passed
> when you run it in your environment? Assuming tests were failing for
> a while, hence a second commit won't do any harm is bad, because it
> makes it difficult to rollback the bad check-in.
I did not get the tests in Felix mode to pass in my environment. I
looked at the stack trace which did not seem to relate to my changes, so
I checked out a fresh workspace to ran QL against that, same error. So I
concluded my changes did not cause the failure and those failures are
probably known issues. Looking back, I agree with you that even a change
don't do any harm will hurt the situation. :-(
- Hong
> I could be wrong here, but I did a quick analysis of hudson log files.
> I see tests started failing from
> http://kohsuke.sfbay/hudson/job/v3-quicklook/5305/ (approx. 6AM this
> morning) onwards. This corresponds to v3 build:
> http://kohsuke.sfbay/hudson/view/GFv3/job/glassfish-v3/4101/. That
> corresponds to following check in:
> http://kohsuke.sfbay/hudson/view/GFv3/job/glassfish-v3/4101/changes#detail0
> (rev 20933 in svn). Now, there have been so many check-ins since then
> that we can't safely rollback the faulty one without having to wait
> for the engr come online.
>
> Thanks,
> Sahoo
>
> Hong Zhang wrote:
>> Hi, Sahoo
>> When I ran QL this morning (about 8am PT, I just checked out a
>> fresh v3 workspace and built from source ), all the QL in HK2 module
>> passed while the QL in Felix mode failed with this message:
>> com.sun.enterprise.module.ResolveError: Failed to start
>> org.glassfish.web:jsf-connector:1.0
>> I suspect these tests were failing for a while. The reason we did
>> not notice it is because the test results for Felix were not
>> previously shown on Hudson. Kohsuke just checked in a change later
>> this morning to show the Felix results.
>>
>> - Hong Sahoo wrote:
>>> Those who have changed code recently are requested to take a look at
>>> hudson QL failures.
>>>
>>> Thanks,
>>> Sahoo
>>>
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail: dev-unsubscribe_at_glassfish.dev.java.net
>>> For additional commands, e-mail: dev-help_at_glassfish.dev.java.net
>>>
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: dev-unsubscribe_at_glassfish.dev.java.net
>> For additional commands, e-mail: dev-help_at_glassfish.dev.java.net
>>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe_at_glassfish.dev.java.net
> For additional commands, e-mail: dev-help_at_glassfish.dev.java.net
>