Lloyd L Chambers wrote:
> Dinesh et al,
>
> I'd like to see it run (and succeed) on a dual or quad-core Mac OS X
> 10.4.8 box for 3 successive 'runtest-ee' runs and 3 subsequent
> 'runtest-ee-standalone'. Not wiping out and redoing it each time, but
> running the test suite 3 times.
> Last night, I had 3 successful runs, then a failure. Does tinderbox
> ever try something like that?
No tinderbox will not try 3 runs on existing workspace, it moves last
workspace to workspace.prior and checkout new one as fresh workspace.
But hudson is using existing appserv-tests module (without wiping out
and recheckout from scratch) with only updated sources all the time, it
works on Solaris sparc, see the console log here:
http://glassfishbuildtools.sun.com:8000/hudson/job/glassfish/1490/console
> It's easy enough to claim something works when the slate is wiped
> clean every time, and each run is fresh. If that is a requirement, so
> be it, but let's say so in the directions.
This harness is not tested for such a robustness. I don't know that much
deeper into quicklook to tell you what may be going wrong.. but if you
want you can create issues for such failures, someone can look at this
later.
thanks
Dinesh
> Lloyd
>
> On Dec 8, 2006, at 6:59 PM, Craig L Russell wrote:
>
>> Hi Dinesh,
>>
>> Just out of curiosity, do you run EEQL on any fast multicore
>> machines like the ones Lloyd is using, before promoting the build?
>>
>> Craig
>>
>> On Dec 8, 2006, at 3:27 PM, Dinesh Patil wrote:
>>
>>> This is not valid issue, EE Quicklook is passing on Hudson: http://
>>> glassfishbuildtools.sun.com:8000/hudson/job/glassfish/
>>>
>>> We promoted the builds yesterday, so this is something to do with
>>> your env, setup, where its failing.
>>>
>>> If you feel there is a bug, file a issue with reproduceable steps
>>> and will take a look.
>>> thanks
>>> Dinesh
>>>
>>> Lloyd L Chambers wrote:
>>>
>>>> Nandini,
>>>>
>>>> No, it does not have to be consistent and always reproducible.
>>>> Have you ever experienced a threading bug?
>>>>
>>>> The bug is real, and a serious impediment to my work. I take
>>>> offense at your suggesting that it doesn't exist. If it works
>>>> for 99/100 developers, great. But it doesn't work for me, on two
>>>> different machines, both Mac OS X, one Power PC and one x86.
>>>> Those are the facts.
>>>>
>>>> Lloyd
>>>>
>>>> On Dec 8, 2006, at 2:56 PM, Nandini Ektare wrote:
>>>>
>>>>>
>>>>> Lloyd L Chambers wrote:
>>>>>
>>>>>> Nandini,
>>>>>>
>>>>>> Failures just start happening. No change in config. No change
>>>>>> in code.
>>>>>>
>>>>>> I don't use proxies.
>>>>>>
>>>>>> Remove the warning? Why? I cannot get EE quicklook to run
>>>>>> reliably (or at all sometimes) AND I'VE FOLLOWED ALL THE
>>>>>> DIRECTIONS.
>>>>>
>>>>>
>>>>> Because the reason for warning is wrong and perhaps there is
>>>>> nothing to be warned about to start with. From your writing
>>>>> below two things can be inferred
>>>>> 1. Config for the QL is wrong => EE QL should fail for everyone
>>>>> And that we know is not true
>>>>> 2. This can happen with a fresh checkout or a later run. => not
>>>>> a reproducible scenario
>>>>> They are contradictory. If such basic config was wrong it has to
>>>>> be reproducible. So it not serving any good intention you had
>>>>> when you added it. It will just create confusion when there is
>>>>> none.
>>>>>
>>>>> ----------------------------------
>>>>> "WARNING WARNING WARNING WARNING WARNING WARNING WARNING WARNING
>>>>> WARNING WARNING
>>>>> QuickLook has a bug which will hang early during the ee tests
>>>>> waiting for an http port that doesn't
>>>>> exist (it typically waits until it times out on port 38080 in
>>>>> sqe- domain, which is the http port
>>>>> found in eeconfig/ee-config.properties. But sqe-domain gets
>>>>> (incorrectly) created with port 8080.
>>>>>
>>>>> This can happen with a fresh checkout or a later run.
>>>>>
>>>>> --Lloyd Chambers, Dec 7, 2006
>>>>> ----------------------------------
>>>>>
>>>>>>
>>>>>> Fix it maybe, but this EE QL failure stuff has cost me many
>>>>>> days of work. The last thing I want to do is remove it and let
>>>>>> some other developer suffer the wasted cycles I have.
>>>>>>
>>>>>> Lloyd
>>>>>>
>>>>>> On Dec 8, 2006, at 2:21 PM, Nandini Ektare wrote:
>>>>>>
>>>>>>> Lloyd L Chambers wrote:
>>>>>>>
>>>>>>>> Nandini,
>>>>>>>>
>>>>>>>> I wish it were true. I've tried the variations. I've removed
>>>>>>>> all domains, and redone 'maven configure-cluster'. Last night
>>>>>>>> I got it to run twice,
>>>>>>>
>>>>>>>
>>>>>>> In between 2nd and 3rd run, do you remember what did you do
>>>>>>> other than no change to code? It is hard to understand why all
>>>>>>> of a sudden the third run failed when it had succeeded twice
>>>>>>> before ... but it will be worth knowing if indeed you had done
>>>>>>> something that caused it to fail... perhaps there is some bug
>>>>>>> lurking out there...
>>>>>>>
>>>>>>> By the way did you do anything with your proxy settings? We
>>>>>>> had a hang in EEQL on Mac yesterday due to changed proxy
>>>>>>> settings on the Mac build machine. Actually the hang is not
>>>>>>> really a hang...the step (verify-setup) that waits to connect
>>>>>>> to the admin url times out after 360000 ms...so its just
>>>>>>> waiting. The actual reason is in earlier steps where the
>>>>>>> instance sqe-server instance fails to start because its sync
>>>>>>> fails (which fails due to a timeout waiting to connect through
>>>>>>> the proxy to DAS). That is why there is no http server
>>>>>>> listening at 38080...
>>>>>>>
>>>>>>> Which also means, I gathered just now, your warning is wrong!
>>>>>>> It is not http port for sqe-domain but for sqe-server. You
>>>>>>> should remove that warning.
>>>>>>>
>>>>>>>> then fail the 3rd time, with no change to the code!
>>>>>>>>
>>>>>>>> asenv.conf shown below
>>>>>>>>
>>>>>>>> AS_ANT="/gf/run/lib/ant"
>>>>>>>> AS_ANT_LIB="/gf/run/lib/ant/lib"
>>>>>>>> AS_WEBSERVICES_LIB="/gf/run/lib"
>>>>>>>> AS_PERL="%PERL_HOME%"
>>>>>>>> AS_NSS="/gf/run/lib"
>>>>>>>> AS_NSS_BIN="/gf/run/lib"
>>>>>>>> AS_IMQ_LIB="/gf/run/imq/lib"
>>>>>>>> AS_IMQ_BIN="/gf/run/imq/bin"
>>>>>>>> AS_CONFIG="/gf/run/config"
>>>>>>>> AS_INSTALL="/gf/run"
>>>>>>>> AS_JAVA="/System/Library/Frameworks/JavaVM.framework/ Versions/
>>>>>>>> 1.5.0/Home"
>>>>>>>> AS_ACC_CONFIG="/gf/run/domains/domain1/config/sun-acc.xml"
>>>>>>>> AS_JHELP="/gf/run/lib"
>>>>>>>> AS_ICU_LIB="/gf/run/lib"
>>>>>>>> AS_DEF_DOMAINS_PATH="/gf/run/domains"
>>>>>>>> AS_JDMK_HOME="/gf/run/lib/SUNWjdmk/5.1"
>>>>>>>> AS_NATIVE_LAUNCHER="false"
>>>>>>>> AS_NATIVE_LAUNCHER_LIB_PREFIX="/jre/lib/i386/client"
>>>>>>>> AS_WEBCONSOLE_LIB="/gf/run/lib"
>>>>>>>> AS_JATO_LIB="/gf/run/lib"
>>>>>>>> AS_HADB="Does not exist."
>>>>>>>> AS_DERBY_INSTALL="/gf/run/javadb"
>>>>>>>> AS_MFWK_HOME="/opt/SUNWmfwk"
>>>>>>>> AS_ADMIN_PROFILE=cluster
>>>>>>>> AS_ADMIN_SECURE=true
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> Lloyd
>>>>>>>>
>>>>>>>> On Dec 8, 2006, at 1:31 PM, Nandini Ektare wrote:
>>>>>>>>
>>>>>>>>>
>>>>>>>>> I think Hong's issue is different from what you had ...she
>>>>>>>>> had rebuilt one module and it was a known module-specific issue
>>>>>>>>>
>>>>>>>>> For you, based on our previous mail conversations, I gather
>>>>>>>>> you had installed using configure-runtime (that puts ina
>>>>>>>>> different profile - 'developer' into asenv.conf) instead of
>>>>>>>>> configure- cluster
>>>>>>>>> When you run EEQL against this install, it creates new
>>>>>>>>> domains with this setting instead of with 'cluster' fetching
>>>>>>>>> the settings from the asenv.conf
>>>>>>>>> And hence DAS is accessible in non-secure mode at http://
>>>>>>>>> localhost:4848 (similarly other port assignments change)
>>>>>>>>> whereas rest of EEQL assumes a secure mode DAS.
>>>>>>>>> Hence NA fails to contact DAS. ...Try flipping that env
>>>>>>>>> setting to 'cluster' or install cluster mode
>>>>>>>>>
>>>>>>>>> That said I do agree this is a bit confusing ... (but I think
>>>>>>>>> I remember seeing mails with plans to get that setting out
>>>>>>>>> of asenv.conf... )
>>>>>>>>> Lloyd L Chambers wrote:
>>>>>>>>>
>>>>>>>>>> I saw the same problem. I also added the following to the
>>>>>>>>>> Quicklook Read Me last night:
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> ______________________________________________________________
>>>>>>>>>> ___ ___________________________________
>>>>>>>>>> WARNING WARNING WARNING WARNING WARNING WARNING WARNING
>>>>>>>>>> WARNING WARNING WARNING
>>>>>>>>>> QuickLook has a bug which will hang early during the ee
>>>>>>>>>> tests waiting for an http port that doesn't
>>>>>>>>>> exist (it typically waits until it times out on port 38080
>>>>>>>>>> in sqe-domain, which is the http port
>>>>>>>>>> found in eeconfig/ee-config.properties. But sqe-domain
>>>>>>>>>> gets (incorrectly) created with port 8080.
>>>>>>>>>>
>>>>>>>>>> This can happen with a fresh checkout or a later run.
>>>>>>>>>>
>>>>>>>>>> --Lloyd Chambers, Dec 7, 2006
>>>>>>>>>> ______________________________________________________________
>>>>>>>>>> ___ ___________________________________
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> On Dec 5, 2006, at 6:59 PM, Hong Zhang wrote:
>>>>>>>>>>
>>>>>>>>>>> How do we run the EE QuickLook tests these days? I first
>>>>>>>>>>> used the command "maven bootstrap configure-cluster" to
>>>>>>>>>>> bootstrap the latest nightly build and configure a cluster
>>>>>>>>>>> profile. Then I ran "maven -o runtest-ee". The PE part ran
>>>>>>>>>>> fine, but when it's to the EE part, I got something like
>>>>>>>>>>> this:
>>>>>>>>>>>
>>>>>>>>>>> Command create-node-agent executed successfully.
>>>>>>>>>>> Command create-instance executed successfully.
>>>>>>>>>>> start-node-agent --interactive=true --passwordfile /home/
>>>>>>>>>>> hzhang/files/workspaces/sjsas91/glassfish/appserv-tests/
>>>>>>>>>>> config/adminpassword.txt --terse=true --user adminuser --
>>>>>>>>>>> syncinstances=false --startinstances=true --verbose=false
>>>>>>>>>>> -- echo=true sqe-agent
>>>>>>>>>>> Redirecting output to /home/hzhang/files/sun/appserver/
>>>>>>>>>>> nodeagents/sqe-agent/agent/logs/server.log
>>>>>>>>>>> Redirecting application output to /home/hzhang/files/sun/
>>>>>>>>>>> appserver/nodeagents/sqe-agent/agent/logs/server.log
>>>>>>>>>>> Enter Admin User:
>>>>>>>>>>> Enter Admin Password:
>>>>>>>>>>> Enter Master Password:
>>>>>>>>>>> Enter Other Password Information (or ctrl-D or ctrl-Z):
>>>>>>>>>>> com.sun.enterprise.config.ConfigException: Error
>>>>>>>>>>> refreshing ConfigContext:/home/hzhang/files/sun/appserver/
>>>>>>>>>>> nodeagents/sqe- agent/sqe-server/config/domain.xml
>>>>>>>>>>> cause: /home/hzhang/files/sun/appserver/nodeagents/sqe-
>>>>>>>>>>> agent/ sqe-server/config/domain.xml (No such file or
>>>>>>>>>>> directory)
>>>>>>>>>>> at
>>>>>>>>>>> com.sun.enterprise.config.impl.ConfigContextImpl.refresh
>>>>>>>>>>> (ConfigContextImpl.java:386)
>>>>>>>>>>> at
>>>>>>>>>>> com.sun.enterprise.config.impl.ConfigContextImpl.refresh
>>>>>>>>>>> (ConfigContextImpl.java:468)
>>>>>>>>>>>
>>>>>>>>>>> So I checked, and there is indeed no /home/hzhang/files/
>>>>>>>>>>> sun/ appserver/nodeagents/sqe-agent/sqe-server/config/
>>>>>>>>>>> domain.xml there. Anything wrong in my setup? Thanks.
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> --------------------------------------------------------------
>>>>>>>>>> --- ----
>>>>>>>>>> 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
>>>>>>>>
>>>>>>>
>>>>>>> -----------------------------------------------------------------
>>>>>>> --- -
>>>>>>> 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
>>>>>
>>>>
>>>> -------------------------------------------------------------------- -
>>>> 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
>>>
>>
>> Craig Russell
>> Architect, Sun Java Enterprise System http://java.sun.com/products/jdo
>> 408 276-5638 mailto:Craig.Russell_at_sun.com
>> P.S. A good JDO? O, Gasp!
>>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe_at_glassfish.dev.java.net
> For additional commands, e-mail: dev-help_at_glassfish.dev.java.net
>