Important: 2 changed files were not checked in. My mistake. It should
be OK now. In the meantime you can stop instances by explicitly giving
the "--froce true" arg.
Funny how QL tests didn't detect the problem...
On 11/2/2010 11:03 AM, Byron Nevins wrote:
> Yes - that is how stop-server commands work. Before the --force
> option was 100% ignored. Now it is used. If you set --force to
> false, then the server will not stop. That's because System.exit() is
> not called and your server has non-daemon threads left running - so it
> will never stop.
>
> See [1] for details
>
> https://glassfish.dev.java.net/issues/show_bug.cgi?id=13951
>
> =============================
>
>
> What this means: The default for --force is true for both types of
> servers. Just stick with the default.
>
>
> On 11/1/2010 5:22 PM, Ken wrote:
>> I don't understand what I am observing with the asadmin stop-instance
>> command (current build as of around noon
>> today, rev 42362). I have an orb failover test that runs a number of
>> requests, then calls (using Runtime.exec)
>> asadmin stop-instance NAME. I have tested this with and without the
>> --force true option. In either case, the command
>> prints out:
>>
>> The instance, NAME, is stopped.
>> Command stop-instance executed successfully.
>>
>> But WITHOUT the --force true option, the server instance PROCESS
>> never dies: it just stays around in an
>> odd partially shutdown state, probably with the application shutdown,
>> causing spurious CORBA errors
>> (basically looks like the EJB does not exist). This causes my
>> failover test to fail.
>>
>> WITH the --force option, everything works correctly: the ORB fails
>> over to another instance in the cluster,
>> and stays on the other instance when I bring the shutdown instance
>> back up with an asadmin start-instance command.
>> ORB failover is dependent on getting a communication failure with the
>> instance in order to detect the failover.
>> I also observe that an instance of the app server is running with a
>> new PID, so apparently in this case
>> the process really was destroyed, which is what I require in the
>> failover test.
>>
>> Is stop-instance supposed to work this way?
>> I remember a discussion about (lack of) daemon threads, is this
>> related to that problem?
>>
>> Thanks,
>>
>> Ken.
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: dev-unsubscribe_at_glassfish.dev.java.net
>> For additional commands, e-mail: dev-help_at_glassfish.dev.java.net
>>
>
--
Byron Nevins - Oracle Corporation
Home: 650-359-1290
Cell: 650-784-4123
Sierra: 209-295-2188