On Sep 3, 2009, at 7:28 PM, 梁昆 wrote:
I think you are using deployment based "farm" clustering to deploy to both A and B at once? This is independent of whether you are using WADI to replicate session state between your cluster members.
I am not very familiar with the deployment based clustering, and I prefer to use plugin based clustering, which however is only available in 2.2 (soon to be released).
(1) seems like a bug to me. I guess you can restart your app in B on the command line or in the admin console, but I think the app should remain started on B.
(2) deployment based clustering works by deploying a synthetic "master" application on the controller server that manages the actual app running on each cluster member. I think that the master app name is derived from the actual app name by appending "_G_MASTER". So, I think you are trying to undeploy the master app from node B, whereas it is running on node A. What happens if you try to undeploy com.ibm.wasce.samples/cviewer/126.96.36.199/car from node B?