Migration failures may occur when migrating application run-time configurations for a large complex network deployment to a higher version of WebSphere Application Server (WAS) Network Deployment (ND) on an AIX platform. The WebSphere migration utilities such as WASPreUpgrade and WASPostUpgrade provide the necessary mechanisms to move the existing configuration information such as older defaults and settings including ports, JVM parameters, etc., from a previous version to a higher version of the WebSphere Application Server. These migration failures may happen for several reasons such as any shortcomings in the procedure or a peculiar deployment topology chosen by the customer or due to any underlying network and synchronization failures from configurations in WebSphere or AIX.

Migration Environment

• Source - WebSphere Application Server ND V6.0
• Target - WebSphere Application Server ND V7.0
• Operating System - 64-bit AIX6.1

Migration Failures due to Shortcomings in the Procedure
Enough file system space has not been allocated to cater to the binaries and profiles

Permission has not been set to read, write or create on the file system

WebSphere Application Server product has not been correctly installed or the profiles have not been properly created for the target environment.

Migration Failures from Peculiar Topology
Peculiar topology may include a managed node with a federated application server (outside the cluster) and co-located deployment manager, and clustered application servers in the cell. The process of migration in the WebSphere product specifies migrating a deployment manager profile and then migrating application servers nodes. However, migration failure may occur whenthe  deployment manager node is migrated prior to the clustered application server nodes.

Migration Failures from WebSphere/AIX Configurations
The migration synchronization process is observed to fail when the system user runs out of file handles with default AIX 'nofiles' limit setting (i.e., ulimit -n)  of 2000. This 'nofiles' value proves to be too low, resulting in errors opening files or establishing connections in the migration process.

It is also known that the migration process synchronizes data in the migrated node with the deployment manager during which the contents of the new profile's configurations are uploaded to the deployment manager, one file at a time.  Some 'network connection reset' or 'network read error’ condition may occur due to migration tool saturating the deployment manager's connections with incoming data if the max connections on the TCP channel configured are too low (say default 100).

Conclusion
These WebSphere migration failures can cause problems for the customers.