Replicated idletimeoutsecs in 5.1 Cluster??

Question

It looks like when <idle-timeout-seconds> is changed in the console for a
          given stateful session bean in a WL 5.1 cluster, that change is applied
          across all nodes in a cluster, even though if one views the value from any
          of the other nodes, it appears unchanged. Is this expected behavior?
          We experienced this about a week ago -- we noticed that our
          idle-timeout-seconds for one of our stateful beans was set too high, so we
          decided to reduce it. Prior to making the change permanent in the
          weblogic-ejb-jar we made it temporarily in the console for one of the nodes
          and figured we'd monitor it. The change automatically applied to all nodes
          in the cluster. When I looked at the other nodes in the console, the idle
          timeout seconds as displayed in the console itself still contained the
          original value, but upon scrolling down and looking at the EJB passivation
          activity, I could see that they were clearing passivating at the point of
          the new reduced idletimeoutseconds value I had set on the one node.
          

Answer 1

We do not have this feature as of now.
          Anand Byrappagari wrote:
          > Rajesh,
          >
          > Sure I can do that but that only allows we to weight the number of calls
          > going to each server but it does not allow me to tweek the servers
          > performace. This is orthogonal to what I want. Let me explain what I mean:
          >
          > Server 1 - weight 50
          > Server 2 - weight 100
          > Using the weighted load balancing I can direct double the amount of traffic
          > to go to Server 2 but that is not all that I want.
          >
          > Server 1 - 1 GB memory
          > Server 2 - 2 GB memory
          > I am interested in tweeking the performance of Server 2 but controlling the
          > number of bean instance allowed. Since Server 2 has a larger memory and may
          > be more CPUs I might want to allow the bean cache to contain twice as many
          > instance or maybe I want the timeout interval to be twice as much since I
          > can afford the extra memory there by gaining some performace improvement by
          > not passivating the beans as often.
          >
          > Requiring all the replicated instance to have the same values seems to be an
          > unnecessary restriction or maybe I am missing something.
          >
          > -- Anand
          >
          > "Rajesh Mirchandani" <[email protected]> wrote in message
          > news:[email protected]...
          > > Anand,
          > >
          > > If you really want to load the balance depending on the capabilities of
          > the
          > > hardware/OS you could use a different load balancing algorithm.
          > >
          > > More info at
          > http://edocs.bea.com/wls/docs61/cluster/features.html#1006780.
          > >
          > >
          > > Anand Byrappagari wrote:
          > >
          > > > I have a similar problem. I want to be able to set different values for
          > > > MaxBeansInCache of a entitybean for different managed servers belonging
          > to a
          > > > weblogic cluster. Is it possible to do this in WL6.1 cluster? Since the
          > only
          > > > way to specify this is in the deployment descriptor, all the targets of
          > this
          > > > deployment will use the new value. Same goes for IdleTimeoutSeconds &
          > > > ReadTimeoutSeconds.
          > > >
          > > > This sounds like a big limitation, if I have two WL instances running on
          > > > different powered machines I would like to have different values
          > assigned to
          > > > MaxBeansInCache to accomodate the difference in the capabilities of the
          > > > machine. Can someone please confirm that this is indeed true. If it is
          > not
          > > > can you please tell me how to set this up.
          > > >
          > > > Thanks,
          > > > Anand
          > > >
          > > > "Michael Joseph" <[email protected]> wrote in message
          > > > news:[email protected]...
          > > > > It looks like when <idle-timeout-seconds> is changed in the console
          > for a
          > > > > given stateful session bean in a WL 5.1 cluster, that change is
          > applied
          > > > > across all nodes in a cluster, even though if one views the value from
          > any
          > > > > of the other nodes, it appears unchanged. Is this expected behavior?
          > > > >
          > > > > We experienced this about a week ago -- we noticed that our
          > > > > idle-timeout-seconds for one of our stateful beans was set too high,
          > so we
          > > > > decided to reduce it. Prior to making the change permanent in the
          > > > > weblogic-ejb-jar we made it temporarily in the console for one of the
          > > > nodes
          > > > > and figured we'd monitor it. The change automatically applied to all
          > > > nodes
          > > > > in the cluster. When I looked at the other nodes in the console, the
          > idle
          > > > > timeout seconds as displayed in the console itself still contained the
          > > > > original value, but upon scrolling down and looking at the EJB
          > passivation
          > > > > activity, I could see that they were clearing passivating at the point
          > of
          > > > > the new reduced idletimeoutseconds value I had set on the one node.
          > > > >
          > > > >
          > > > >
          > >
          > > --
          > > Rajesh Mirchandani
          > > Developer Relations Engineer
          > > BEA Support
          > >
          > >
          Rajesh Mirchandani
          Developer Relations Engineer
          BEA Support