I have a cluster setup as follows
Two weblogic instance on seperate boxes in cluster hosting Presentation logic and two weblogic instance in cluster hosting Object layer(Ejbs). When i remove one box hosting presention and object layer from the network, automatic failover takes place but it is too slow. What may be the problem, how can i fix it.
Thanks
rajiv
-
Automatic failover in cluster (5 messages)
- Posted by: Rajiv Mulay
- Posted on: November 16 2001 16:59 EST
Threaded Messages (5)
- Automatic failover in cluster by Pratap Das on November 17 2001 02:08 EST
- Automatic failover in cluster by Doc Pisapati on November 19 2001 10:21 EST
- Automatic failover in cluster by Rajiv Mulay on November 19 2001 13:34 EST
-
Automatic failover in cluster by Pratap Das on November 19 2001 10:41 EST
- Automatic failover in cluster by Rajiv Mulay on November 22 2001 02:41 EST
-
Automatic failover in cluster by Pratap Das on November 19 2001 10:41 EST
- Automatic failover in cluster by Rajiv Mulay on November 19 2001 13:34 EST
-
Automatic failover in cluster[ Go to top ]
- Posted by: Pratap Das
- Posted on: November 17 2001 02:08 EST
- in response to Rajiv Mulay
Could you clarify your architecture - do you have 4 weblogic instances on 4 machines or 2 instances on 2 machines?
--Das -
Automatic failover in cluster[ Go to top ]
- Posted by: Doc Pisapati
- Posted on: November 19 2001 10:21 EST
- in response to Rajiv Mulay
Rajiv, could you tell me what OS and hardware you are running? If Win2K, are you using NLB for presentation (IIS or Apache)? If Solaris, are you using "Full Moon"? -
Automatic failover in cluster[ Go to top ]
- Posted by: Rajiv Mulay
- Posted on: November 19 2001 13:34 EST
- in response to Doc Pisapati
Hi,
I'm having 4 instances on two machines, the OS is solaris/sparc
-
Automatic failover in cluster[ Go to top ]
- Posted by: Pratap Das
- Posted on: November 19 2001 22:41 EST
- in response to Rajiv Mulay
So, I presume you are trying out the Multi-Tier Architecture described in BEA's http://e-docs.bea.com/wls/docs61/cluster/planning.html#1094155 document.
Here is a snippet from their docs - maybe you should go thru it.
<snippet>
Limitations of Multi-Tier Architecture
Because the advanced configuration cannot optimize object calls using the collocation strategy, the Web application incurs network overhead for all method calls to clustered objects. This overhead may be acceptable, however, if your Web application requires any of the benefits described in Benefits of Multi-Tier Architecture.
</snippet>
HTH
--Das -
Automatic failover in cluster[ Go to top ]
- Posted by: Rajiv Mulay
- Posted on: November 22 2001 14:41 EST
- in response to Pratap Das
I was not reffering to application overhead due to method calls. The failover takes more time why is it so