Hallo,

what are your thoughts / ideas on the following scenario:
An aggregator running with a jdbc-aggregation-repository against a shared 
database in  a jboss-cluster. The aggregator has a completionPredicate and a 
completionTimeout. The aggregator is used to correlate responses with requests 
and have a response-timeout. Timeouts are not ignored, but further processed.

Currently, I can theoretically come up with one problem: recovery happens on 
both machines, thus timeouts can happen on both nodes "simultaneously" and get 
further processed. This will get the system running into a collision, as the 
processing behind the aggregator is _not_ idempotent. Duplicates are sorted out 
before the aggregator.

What are your thoughts / ideas on this?

-Sebastian


-- 
Empfehlen Sie GMX DSL Ihren Freunden und Bekannten und wir
belohnen Sie mit bis zu 50,- Euro! https://freundschaftswerbung.gmx.de

Reply via email to