P.S.: also a test case reproducing this issue will be very helpful. You could attach it to the jira ticket, so I could also check how you're doing the marshalling/unmarshalling and figure out if this problem happens in the most general case or you just found an edge case.
thanks for having reported this. What you found is quite interesting and I will give a look at this issue asap.
In the meanwhile it would be great if you could open a ticket about it here: https://issues.jboss.org/browse/DROOLS
Thanks again and regards,
MarioOn Mon, Feb 11, 2013 at 10:56 AM, Magnus Vojbacke <firstname.lastname@example.org> wrote:
I run into memory issues when using marshaling or unmarshalling Drools stateful sessions.
I was going to send this to rules-users, but I think the content is too technical and implementation oriented to fit into that list.
It seems that the memory consumption increases after a session has been unmarshalled. That is to say: an unmarshalled session of size x uses much more memory than if I had started with an empty session and inserted x number of facts.
Digging into a memory profile, I found that after unmarshalling, my ReteooStatefulSession instance holds on to a MarshallerReaderContext instance. The MarshallerReaderContext consumes a lot of memory.
Digging into the unmarshalling code, it seems that the MarshallerReaderContext is a utility class used during unmarshalling of sessions. Intuitively, it seems wrong that a reference to this object should be held after unmarshalling is finished.
Is this a bug in Drools? Would it be ok if I filed this as an issue at the Drools issue repository?
rules-dev mailing list