[jbosscache-dev] Notifications during state transfer WAS: [JBossCache] - Replication and notifications

Vladimir Blagojevic vladimir.blagojevic at redhat.com
Tue Nov 13 10:25:12 EST 2007


Brian Stansberry wrote:
> I'm sure this is discussed on a thread if someone wants to search. :)
>
> I'm lazy so I'll just guess about the reason. Going through the 
> interceptor chain adds all sorts of possibly unwanted side effects; 
> e.g. creation of in-memory nodes cause writes to persistent store, 
> replication, gravitation, etc.  Many can be eliminated by setting 
> options, but thinking through all the ways it can break will take a 
> lot of work and it will inherently be fragile.
>
> Manik's right though; my question was about the Node Modified bit. :)
>

Yeah I think there was an overhead of locking all these nodes or 
something along these lines.
Node Modified bit - I am not sure why we did not do it. Maybe as you say 
- we forgot.





More information about the jbosscache-dev mailing list