No, that was the admin.output file created by the test harness. With verbose=true, the upgrade output goes to std out and so is in the file. I don't have the server log to attach (the domain is removed at the end of the test).
Did you think I had AS_LOGFILE set? Ok, I can turn verbose off and set that. I've attached it, but the output is the same except for the exit code, which is what I was asking about in the first place. Also, the AS_LOGFILE doesn't contain the information about the timeouts.
If this still isn't the same file, let me know. I don't know what other info I can get you though. Command told me it didn't work, but succeeded anyway.
Bobby
On Nov 16, 2010, at 2:32 PM, Byron Nevins wrote:
> The attachment is the SERVER log -- not the CLI log...
>
>
> On 11/16/2010 8:44 AM, Bobby Bissett wrote:
>>
>> Sure, it's attached. The start instance problem has now been fixed, so I reproduced this with b29. So I don't need anything, but if the "start cluster passes even though instances failed" issue is a bug, the info is attached.
>>
>> The part I was asking about is:
>> [java] asadmin --host localhost --port 4848 --user admin --passwordfile /Users/bobby/work/ws/appserv-tests/devtests/admin/cli/resources/uppassword.txt --interactive=false --echo=true --terse=true start-cluster --verbose=false upcluster
>> [java] upin2: Timed out waiting for upin2 to start.
>> [java] upin1: Timed out waiting for upin1 to start.
>> [java]
>> [java] The command start-instance failed for: upin2 upin1
>>
>> Just wanted to see if asadmin should be returning true in this case.
>>
>> On Nov 15, 2010, at 8:11 PM, Byron Nevins wrote:
>>
>>> Can you share the contents of the cli logfile after the run?
>>>
>>>
>>> On 11/15/2010 12:56 PM, Bobby Bissett wrote:
>>>> Hi all,
>>>>
>>>> I just committed a cluster upgrade test to the admin dev tests. It's commented out of the cli target because it can't pass without some issues being fixed. But it can be run with the param: -Dteststorun=cluster-upgrade
>>>>
>>>> Question 1:
>>>> The instances do not come up all the way due to several startup errors. The output is:
>>>>
>>>> upin1: Timed out waiting for upin1 to start.
>>>> upin2: Timed out waiting for upin2 to start.
>>>> The command start-instance failed for: upin1 upin2
>>>>
>>>> ...and yet the test passes (the asadmin command returns true). Is that expected?
>>>>
>>>> Question 2:
>>>> Because of the timeouts, the instances are semi-running (there is a Java process and they're sending GMS announcements that they're up), but can't be stopped with stop-cluster. So after the test exits there are still 4 Java processes running for the the two instances. Is there something else I can do in the test to clean this up?
>>>>
>>>> If someone is looking for a way to test the robustness of start/stop cluster, here's a good test case for you (until everything is working normally again)
>>>>
>>>> Admin test info again: http://wikis.sun.com/display/GlassFish/AdminTests
>>>> See the comments in ClusterUpgradeTest.java for full details.
>>>>
>>>> Thanks,
>>>> Bobby
>>>> [...]
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: admin-unsubscribe_at_glassfish.dev.java.net
>> For additional commands, e-mail: admin-help_at_glassfish.dev.java.net
>
> --
> Byron Nevins - Oracle Corporation
> Home: 650-359-1290
> Cell: 650-784-4123
> Sierra: 209-295-2188
> --------------------------------------------------------------------- To unsubscribe, e-mail: admin-unsubscribe_at_glassfish.dev.java.net For additional commands, e-mail: admin-help_at_glassfish.dev.java.net