Hi,
On 12.11.10 16:27, "Bengt Rodehav" <[email protected]> wrote: >Clement, > >Now I've tested trunk again and I can't seem to get it to work. When the >exception is thrown in the start() method, the instance disappears from >the >iPOJO tab in Webconsole. I assume it is because it is STOPPED. > >When I change my configuration (so that it's correct) I'm hoping the >instance will start again but nothing happens. > >Does your test case involve the configuration admin? Remember that my >controller property is connected to the configuration admin. This makes it >possible for me to enable/disable my service with a configuration >property. >Could this lead to timing issues? Will the instance receive configuration >changes while it is stopped? It is directly using the 'reconfigure' method... So you're right. I should try with the config admin. I will look into that this weekend, and keep you posted. Clement > >BTW the Svn revision of iPOJO that I've tested is 1033833. > >/Bengt > >2010/11/12 Bengt Rodehav <[email protected]> > >> Clement, >> >> I've tested the workaround again to get a feel for what is happening. >> >> I use the same code I wrote in my previous post. When forcing an >>exception >> in the start() method, the following situation occurs: When looking at >>the >> iPOJO instance via the Webconsole, it is still marked as valid but the >> architecture says the valid property is false. >> >> Looking at the configuration tab in the Webconsole, it seems the valid >> property (according to configuration admin) is still true. It seems >>like I >> cannot change the valid property (mValid field) directly like I do. Do I >> have to use configuration admin? >> >> /Bengt >> >> 2010/11/12 Clement Escoffier <[email protected]> >> >> Hi, >>> >>> >>> On 11.11.10 09:51, "Bengt Rodehav" <[email protected]> wrote: >>> >>> >Clement, >>> > >>> >Unfortunately it's still not working for me. I firs tried the >>>workaround >>> >which basically goes as follows: >>> > >>> > @Controller >>> > @Property(name = "enable", mandatory = true) >>> > private boolean mValid; >>> > . >>> > . >>> > . >>> > try { >>> > // start the route >>> > } catch (Exception e) { >>> > mValid = false; >>> > } >>> > >>> >If an exception is thrown, then the instance is made invalid. >>>However, I >>> >cannot get it valid again even if I fix the underlying configuration >>> >problems and enable it again. >>> > >>> >I get the same result when using the trunk version of iPOJO. >>> >>> Hum, that's not normal. Check that another exception is not thrown (the >>> log should contain it). >>> >>> > >>> >Have you tested that it's possible to validate a stopped instance this >>> >way? >>> >I'm not sure if I'm doing anything wrong here... >>> >>> Yes, I've implemented a test case, and the instance is restarted. >>> >>> > >>> >I use the following annotations on my class: >>> > >>> >@Component(name = "myName", propagation = true, immediate = true) >>> >@Provides(specifications = myInterface.class, strategy = "SERVICE") >>> >>> I will test with this configuration. >>> >>> > >>> >I usually use the Felix Webconsole with the iPOJO plugin to inspect my >>> >iPOJO >>> >instances. Currently when instances are stopped, they are not shown in >>> the >>> >Webconsole at all while if they are invalidated they are shown with >>>the >>> >state "invalid" and also information about why the instance is not >>>valid. >>> >>> A stopped instance is really stopped. The Architecture service is no >>>more >>> exposed as no data can be collected (all handlers will say : STOPPED). >>> >>> > >>> >It would be very useful if even stopped instances were shown in the >>> >Webconsole and perhaps with some information about why the instance is >>> >stopped (e g the exception thrown during initialization). >>> >>> I've received a similar request, and I'm looking for the best way to >>> implement this. Right now, the log contain the error. >>> >>> Clement >>> >>> > >>> >/Bengt >>> > >>> > >>> > >>> >2010/11/11 Bengt Rodehav <[email protected]> >>> > >>> >> Wow, your're fast... >>> >> >>> >> I'll try both the workaround and the fix since I'll probably need to >>> use >>> >> the workaround until the next iPOJO release. >>> >> >>> >> /Bengt >>> >> >>> >> 2010/11/10 Clement Escoffier <[email protected]> >>> >> >>> >>> >>> >>> >>> >>> On 10.11.10 20:45, "Bengt Rodehav" <[email protected]> wrote: >>> >>> >>> >>> >Thanks a lot Clement. I'll try your workaround while wating for a >>>fix >>> >>>on >>> >>> >trunk. >>> >>> >>> >>> Already Fixed in trunk :-) >>> >>> >>> >>> Clement >>> >>> >>> >>> > >>> >>> >/Bengt >>> >>> > >>> >>> >2010/11/10 Clement Escoffier <[email protected]> >>> >>> > >>> >>> >> Hi, >>> >>> >> >>> >>> >> So, I've checked. So, what I explained works correctly. So if >>>the >>> >>> >>instance >>> >>> >> is INVALID and we reconfigure the instance, the instance state >>>is >>> >>> >> recomputed. However... In your case you're not INVALID but >>>STOPPED. >>> >>> >> Indeed, when the constructor or a @Validate callback throw an >>> >>> exception, >>> >>> >> the instance is shutdown. >>> >>> >> >>> >>> >> Unfortunately, when a stopped instance is reconfigured, the >>>state >>> is >>> >>> not >>> >>> >> recomputed. That's definitely a bug (and a pretty nice catch !). >>> The >>> >>> >> instance should be restarted with the new configuration. This >>>issue >>> >>> will >>> >>> >> be fixed pretty soon in the trunk. >>> >>> >> >>> >>> >> What you can do for now is to use a lifecycle controller >>> >>>(@Controller), >>> >>> >> catch the Exception and set the controller to false. Then the >>> >>>instance >>> >>> >>is >>> >>> >> invalid not stopped. >>> >>> >> >>> >>> >> Regards, >>> >>> >> >>> >>> >> Clement >>> >>> >> >>> >>> >> >>> >>> >> On 10.11.10 18:00, "Bengt Rodehav" <[email protected]> wrote: >>> >>> >> >>> >>> >> >Thanks Clement. >>> >>> >> > >>> >>> >> >I forgot to write that I use iPOJO 1.6.6. >>> >>> >> > >>> >>> >> >/Bengt >>> >>> >> > >>> >>> >> >2010/11/10 Clement Escoffier <[email protected]> >>> >>> >> > >>> >>> >> >> Hi, >>> >>> >> >> >>> >>> >> >> >>> >>> >> >> On 10.11.10 17:04, "Bengt Rodehav" <[email protected]> wrote: >>> >>> >> >> >>> >>> >> >> >I create iPOJO instances from factory configurations using >>> >>> >> >>Configuration >>> >>> >> >> >Admin and File Install. When the iPOJO instance becomes >>>valid I >>> >>>do >>> >>> >>my >>> >>> >> >> >initialisation (which normally starts a Camel route) and >>>when >>> >>>the >>> >>> >>iPOJO >>> >>> >> >> >instance becomes invalid I terminate my processing (which >>>means >>> >>> >>that I >>> >>> >> >> >normally stop a Camel route). >>> >>> >> >> > >>> >>> >> >> > @Validate >>> >>> >> >> > public void start() { >>> >>> >> >> > // Start the route >>> >>> >> >> > } >>> >>> >> >> > >>> >>> >> >> > @Invalidate >>> >>> >> >> > public void stop() { >>> >>> >> >> > // Stop the route >>> >>> >> >> > } >>> >>> >> >> > >>> >>> >> >> >To make it possible to control my iPOJO instances I use a >>> >>> controller >>> >>> >> >> >property as follows: >>> >>> >> >> > >>> >>> >> >> > @Controller >>> >>> >> >> > @Property(name = "enable", mandatory = true) >>> >>> >> >> > private boolean mValid; >>> >>> >> >> > >>> >>> >> >> >This allows me to enable/disable my service via >>>configuration. >>> >>> >> >>However, I >>> >>> >> >> >have noticed that if an exception is thrown in the start() >>> >>>method >>> >>> >> >>above, >>> >>> >> >> >then the iPOJO becomes invalid and there is no way for me to >>> >>>make >>> >>> it >>> >>> >> >>valid >>> >>> >> >> >by changing any configuration property. It seems like I >>>have to >>> >>> >>delete >>> >>> >> >>the >>> >>> >> >> >configuration and create a new one. Have I understood this >>> >>> >>correctly? >>> >>> >> >> >>> >>> >> >> Modifying the configuration should trigger a reconfiguration >>>of >>> >>>the >>> >>> >> >> instance. It should then recompute the state (in theory it >>> should >>> >>> >>work). >>> >>> >> >> Deleting the configuration disposes the instance and >>>recreates a >>> >>>new >>> >>> >> >>one. >>> >>> >> >> >>> >>> >> >> > >>> >>> >> >> >I would like a convenient way to either make my iPOJO >>>instance >>> >>> valid >>> >>> >> >>again >>> >>> >> >> >or a way to dispose of the old instance and create a new >>>one. >>> >>>How >>> >>> >>can I >>> >>> >> >> >accomplish this? It would be convenient if I could flag my >>> iPOJO >>> >>> >>class >>> >>> >> >> >with >>> >>> >> >> >an annotation like "deleteOnException" or something like >>>that. >>> >>> >> >> >>> >>> >> >> Modifying the configuration should do the work. I will check >>> >>>that. >>> >>> >> >> >>> >>> >> >> Regards, >>> >>> >> >> >>> >>> >> >> Clement >>> >>> >> >> >>> >>> >> >> > >>> >>> >> >> >/Bengt >>> >>> >> >> >>> >>> >> >> >>> >>> >> >> >>> >>> >> >> >>> >>> >>>--------------------------------------------------------------------- >>> >>> >> >> To unsubscribe, e-mail: [email protected] >>> >>> >> >> For additional commands, e-mail: [email protected] >>> >>> >> >> >>> >>> >> >> >>> >>> >> >>> >>> >> >>> >>> >> >>> >>> >> >>> >>>>>>--------------------------------------------------------------------- >>> >>> >> To unsubscribe, e-mail: [email protected] >>> >>> >> For additional commands, e-mail: [email protected] >>> >>> >> >>> >>> >> >>> >>> >>> >>> >>> >>> >>> >>> >>>--------------------------------------------------------------------- >>> >>> To unsubscribe, e-mail: [email protected] >>> >>> For additional commands, e-mail: [email protected] >>> >>> >>> >>> >>> >> >>> >>> >>> >>> --------------------------------------------------------------------- >>> To unsubscribe, e-mail: [email protected] >>> For additional commands, e-mail: [email protected] >>> >>> >> --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
