This could lead to problems when a timeout occurs, but it would only throw an exception from the timout thread, so I don't think it would be harmful. However, if you want to cluster your aggregator, you need to implement a clusterable TimerManager (using quartz and a distributed database i guess).
On Thu, May 8, 2008 at 11:07 AM, Ron Gavlin <[EMAIL PROTECTED]> wrote: > What are the implications of using AbstractAggregator with JMSFlow in a > cluster given that its closedAggregators data structure does not appear to be > clusterable? > > - Ron > > -- Cheers, Guillaume Nodet ------------------------ Blog: http://gnodet.blogspot.com/
