Re: Why does sfTerminateWith only throwRemoteException?

On 6/21/06, Goldsack, Patrick <[email protected]> wrote:
>
>
> The problem you will have is that there is no-one listening for that
> exception... termination does not assume that it is being called from
> "above" as this may have already been terminated or indeed may have caused
> the termination by failing.  Indeed, we do not even need to have the Remote
> exception as the method is never called remotely. It is a known issue of the
> interface and is only there because this method was included into the Prim
> interface which is remotable.
>
> If you need to track problems with termination you will need to do this some
> other way - reporting to some function in the system during termination, for
> example, or adding another method to terminate VMs which return exceptions
> to the caller and make this independant of the component termination. Most
> system lifecycles we use are not necessarily tightly bound to that of the
> components which implement it.
>
> Patrick

Yeah, that's what we figured out eventually. We're now logging any
failures that happen
during component termination, and just going on with the method,
masking the exception.
This makes sense anyway, as there's no way to recover from a failed
termination without involving human interaction.

Thanks for the prompt reply,
-- 
Olivier Pernet
Summer student - CERN Openlab

We are the knights who say
echo '16i[q]sa[ln0=aln100%Pln100/snlbx]sbA0D4D465452snlbxq'|dc