Re: POA Destruction problems...

Hi Dion,
I guess I was taking the Henning and Vinoski advice that 
OBJECT_NOT_EXIST should be treated as a definitive "object death 
certificate" a little too seriously.  Nonetheless, thanks for the advise 
- judging from what you said, it looks like the best way to go would be 
to transition the POA Manager to INACTIVE /and then/ destroy the POA. 
 Atleast this way, OBJECT_NOT_EXIST will never get raised to the client 
and the client will not assume the object is permanently gone even if it 
does occasionally need to recover from a TRANSIENT or COMM_FAILURE 
exception.
Much thanks,
    ~Andy

Dion Picco wrote:

>Hi Andy,
>
>On Tue, Sep 14, 2004 at 02:26:32PM -0700, Andy Beard wrote:
>  
>
>>Hi,
>>I'm experiencing some confusion regarding the proper destruction of 
>>PERSISTENT POAs.  The main problem is that we routinely shutdown and 
>>restart applications using persistent POAs in order to integrate 
>>implementation changes.  The question I have is whether or not I should 
>>call destroy() on these persistent POAs prior to shutdown.  We 
>>definitely do not wish to invalidate any client side references to 
>>servants incarnated in these POAs, however if I call destroy on the 
>>encompassing POA, OBJECT_NOT_EXIST exceptions are raised to the client 
>>when invoking requests prior to process termination but post POA 
>>destruction.  This is confusing because when the client is restarted, a 
>>new persistent POA is instantiated and used to incarnate the CORBA 
>>object once again (and subsequently OBJECT_NOT_EXIST is no longer 
>>raised), however the client has been misled into thinking the object is 
>>permanently invalidated.  It seems to me that OBJECT_NOT_EXIST should 
>>not be raised when destroying persistent POAs unless the object is 
>>explicitly deactivated as well.  Just because the POA was destroyed 
>>doesn't mean the object itself was destroyed as well,  Correct?  As a 
>>possible solution, I could simply transition the POA Manager to INACTIVE 
>>when exiting my applications and never explicitly destroy persistent 
>>POAs.  Is this a viable solution?
>>
>>Thanks ahead of time,
>>
>>    ~Andy
>>    
>>
>
>If you have destroyed your POAs, your client application will receive
>OBJECT_NOT_EXIST exceptions in response to requests.  This is the
>behaviour outlined in section 11.2.6 of the CORBA specification (second
>paragraph).  If the ORB cannot find the correct POA, it cannot deliver
>the request to that POA and hence as far as the client is concerned, the
>object doesn't exist.
>
>NOTE: Even though the servant continues to exist within
>the process space of your server, there isn't a POA to deliver the
>requests; hence the ORB cannot deliver the request to an object and it
>assumes that it doesn't exist.
>
>If you are fully shutting down your server, I would expect you to
>receive a TRANSIENT exception since there isn't anything listening for
>requests (POAManagers).  However, if only the POAs are destroyed, then
>the OBJECT_NOT_EXIST is the exception you will receive.  If you are
>destroying and recreating POAs only (instead of restarting servers), you
>should transition your POAManagers into a holding or discarding state
>perhaps.  Or, like you mention, you are receiving requests between POA
>destruction and process termination, you should first transition the
>POAManager to a discarding/holding state before destroying the POAs.  
>
>Cheers.
>
>  
>

_______________________________________________
OB-Users Mailing List - [email protected]
http://mail.ooc.nf.ca/mailman/listinfo/ob-users
Visit our support FAQ before you send a message.
http://www.orbacus.com/faq/support.html