I have a situation where a web application targeted to one server in a cluster causes a deployment error on other servers of the cluster.
I had an original application A (an EAR) deployed on all servers in a cluster. Now I have to deploy another web application B on the same domain on this cluster. However I want that this application be deployed only on one server of the cluster. And when I deploy the app through the console, weblogic does indeed provide me with such an option (deploy to only a specific machine in the cluster). I happily choose that option, finish the deployment and on restart of the server, both the apps - A and B are indeed deployed and working well - so far so good.
The problem arises when I restart the other machines of the cluster. On startup, the other servers complain that it cannot find the path to the application B and fails all deployments (even app A is not deployed). It cannot find the path to application B because I have not copied the build over to the other server. I have not targeted the deployment of application B to other servers and hence did not deem it necessary to copy over the build to other servers. However the server thinks otherwise and refuses to deploy any app :)
Needless to say, when I copy over the build (or even provide an empty web app), the other servers startup and do not attempt to deply application B. It looks like they just expect the build for application B to be available. Does anyone know if this is expected behavior and why? I have searched weblogic documentation and googled without any luck.
Thanks for your response.
NoStage initially. But we have since switched to stage so that the admin server would copy over the archive to other servers. But that's exactly my question - why do the other servers even need this archive?
No, the app does not use any shared application libraries.
This is how staging woks when you use default recommended staging then it will create copy in Admin Server if you select external staging then it will copy single copy on every server to have backup if there is any issues in case of the network failure etc.
If you are more confident then you can work with NoStage mode of deployment.
Kalyan, thanks for your response. I understand the different modes of deployment.
What I dont is why would the the other servers in the cluster look for an archive that is not targeted for deployment on them ?