I agree that a sequence should survive a crash (if using persistent
storage), but I don't think this applies to a normal shutdown. If you're
doing a normal shutdown of the client it means you've finished whatever
you're doing. The server should be told that you're done by sending the
TerminateSequence, otherwise it's going to keep resources tied up with
its RM handling.
- Dennis
On 09/02/2013 07:57 PM, Aki Yoshida wrote:
Hi Dennis,
You mean you want to terminate the active sequence when the endpoint
is shutdown? Maybe I am getting what you meant. A sequence should
survive a crash or a normal endpoint shutdown because the persisted
messages for that ws-rm endpoint must survive. So I don't think we
should change the default setting.
We can already try to avoid having non-terminated sequences by setting
the sequence length or validity and let the ws-rm runtime terminate
the sequences periodically.
regards, aki
2013/9/1 Dennis Sosnoski <[email protected]>:
Right now WS-RM handling allows the user to configure sequence termination
on shutdown, but defaults to no termination of sequences. This looks like an
incorrect default, since WS-RM relies on having a TerminateSequence message
in order to free up resources at the destination. I'd like to change this in
trunk to default to sending the termination, while allowing the user to
override with an explicit setting.
The cost of this is an added message exchange (or potentially two, if the
server also terminates a reverse sequence) as part of the client shutdown
process.
Anyone object?
- Dennis
--
Dennis M. Sosnoski
Java SOA and Web Services Consulting <http://www.sosnoski.com/consult.html>
CXF and Web Services Security Training
<http://www.sosnoski.com/training.html>
Web Services Jump-Start <http://www.sosnoski.com/jumpstart.html>
