Re: [JBoss-user] clustering question
I know this is OT, but the thread reminded me of this... Has anyone looked at using something like "JSR-107: JCACHE - Java Temporary Caching API" to help in keeping cached entity beans in a cluster synchronized instead of relying solely on the database? David Jencks wrote: >I don't see a response yet from the experts...so I will jump in. > >Don't use commit option A with clustering, and do use whatever your >database supplies for select for update. The clustering implemention >relies on database locking to synchronize the different copies of entities. > >I actually don't know of any other way to get consistency, but would be >interested in finding out if there are other ways. > >david jencks > ___ JBoss-user mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/jboss-user
RE: [JBoss-user] clustering question
> I don't see a response yet from the experts...so I will jump in. Sometimes, I sleep ;) > Don't use commit option A with clustering, and do use whatever your > database supplies for select for update. The clustering implemention > relies on database locking to synchronize the different copies of > entities. exact. See specific case for read-only beans. > I actually don't know of any other way to get consistency, but would be > interested in finding out if there are other ways. Distributed cache is one way, but you already knew that :) Cheers, Sacha ___ JBoss-user mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/jboss-user
RE: [JBoss-user] clustering question
Hello Dennis, For now, for entity beans, we use pessimistic locking at the database level i.e. we use the ~"SELECT ... FOR UPDATE" features of databases to lock database rows. Consequently, you need to run in commit option C (or B if you have a kid whose name starts with a B... it won't make a huge difference.) Now, if you use commit option A, it means that you have read-only beans. In this case, it makes perfect sense. Then, if you want caching + commit option C, then you will either have to: 1°) wait on a future release of JBoss that may have distributed cache (nothing planed) 2°) use one of JBoss' partners distributed cache (see the partner page on JBoss web site) Anyway, read the clustering doco, you will understand all this very easily ;) Cheers, Sacha > Let's say I cluster two jboss servers and I have an Entity Bean with > commit option A. > > Are the attributes of the Entity Bean clustered? ie: user on server 1 > calls a set Method, then user on server 2 calls the get method for that > attribute. Can I expect the attribute to be current on server 2? > > If the data wasn't clustered I would say no because server 2 would think > it has exclusive access to the data (commit option A) so it wouldn't > load it. I want to know if the clustering broadcasts the attribute > changes?? > > Thanks > Dennis > > > ___ JBoss-user mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/jboss-user
Re: [JBoss-user] clustering question
I don't see a response yet from the experts...so I will jump in. Don't use commit option A with clustering, and do use whatever your database supplies for select for update. The clustering implemention relies on database locking to synchronize the different copies of entities. I actually don't know of any other way to get consistency, but would be interested in finding out if there are other ways. david jencks On 2002.04.29 17:14:51 -0400 Dennis Muhlestein wrote: > Let's say I cluster two jboss servers and I have an Entity Bean with > commit option A. > > Are the attributes of the Entity Bean clustered? ie: user on server 1 > calls a set Method, then user on server 2 calls the get method for that > attribute. Can I expect the attribute to be current on server 2? > > If the data wasn't clustered I would say no because server 2 would think > it has exclusive access to the data (commit option A) so it wouldn't > load it. I want to know if the clustering broadcasts the attribute > changes?? > > Thanks > Dennis > > > ___ JBoss-user mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/jboss-user
[JBoss-user] clustering question
Let's say I cluster two jboss servers and I have an Entity Bean with commit option A. Are the attributes of the Entity Bean clustered? ie: user on server 1 calls a set Method, then user on server 2 calls the get method for that attribute. Can I expect the attribute to be current on server 2? If the data wasn't clustered I would say no because server 2 would think it has exclusive access to the data (commit option A) so it wouldn't load it. I want to know if the clustering broadcasts the attribute changes?? Thanks Dennis signature.asc Description: This is a digitally signed message part