Sounds reasonable to me to, however I'm wondering what is happening with
the false return value. Wouldn't that result in a similar error (no
route found). If the up-bubbling exception (RETIRED_CAUSE_CODE) leads to
the same situation but with a more specific error message, that's fine
(my concern is just whether this is the best possible position for the
check).
Tammo
On 22.10.2010 05:12, Jeff Yu wrote:
> Removing the isActive() check makes sense to me.
>
> Regards
> Jeff
>
> On Wed, Oct 20, 2010 at 11:42 PM, Milinda Pathirage <
> milinda.pathir...@gmail.com> wrote:
>
>> I had a offline chat with Tammo and found out that this happens due to
>> isActive check in line 299 of BpelProcess.java.
>>
>> boolean invokeProcess(final MyRoleMessageExchangeImpl mex, boolean
>> enqueue) {
>> return invokeProcess(mex, new InvokeHandler() {
>> public boolean invoke(PartnerLinkMyRoleImpl target,
>> PartnerLinkMyRoleImpl.RoutingInfo routing, boolean createInstance) {
>> if (routing.messageRoute == null && createInstance &&
>> isActive()) {
>> // No route but we can create a new instance
>> target.invokeNewInstance(mex, routing);
>> return true;
>> } else if (routing.messageRoute != null) {
>> // Found a route, hitting it
>>
>>
>>
>> _engine.acquireInstanceLock(routing.messageRoute.getTargetInstance().getInstanceId());
>> target.invokeInstance(mex, routing);
>> return true;
>> }
>> return false;
>> }
>> }, enqueue);
>> }
>>
>> Due to this check process state checking in invokeNewInstance will never
>> get
>> executed.
>>
>> if (_process._pconf.getState() == ProcessState.RETIRED) {
>> throw new InvalidProcessException("Process is retired.",
>> InvalidProcessException.RETIRED_CAUSE_CODE);
>> }
>>
>> Is it ok to remove this isActive check or will there be any issues if I
>> remove this?
>>
>> Thanks
>> Milinda
>>
>>
>>
>> On Wed, Oct 20, 2010 at 5:10 PM, Milinda Pathirage <
>> milinda.pathir...@gmail.com> wrote:
>>
>>> Hi Devs,
>>>
>>> Current way of handling instance creation requests to retired processes
>>> doesn't throw any error to client and client will get a timeout when mex
>>> timeout happens. And the request will be queued in the job processor. Is
>>> this the correct behavior. Don't we need to send a error back to the
>> client
>>> saying the process is retired.
>>>
>>> Thanks,
>>> Milinda
>>>
>>>
>>
>
>
>
--
Tammo van Lessen - http://www.taval.de
