I've observed different behaviors when externalizing HTTP sessions to RHDG in JBoss EAP 8.0, depending on the configuration method used.

Observation:
When using the currently recommended configuration method for HTTP session externalization (as described in Red Hat JBoss Enterprise Application Platform 8.0 - Configuration Guide)), if even one of the configured RHDG nodes is not running when EAP starts, the web application fails to deploy. This appears to be due to EAP attempting to connect to the stopped RHDG node, resulting in a connection failure.

Comparison with Legacy Method:
In contrast, when using the legacy HTTP session externalization method (as described in Red Hat JBoss Enterprise Application Platform 8.0 - Configuration Guide), the web application on EAP successfully starts and operates normally as long as at least one RHDG node is running.

This discrepancy could significantly impact availability and operational efficiency in production environments. It's desirable for web applications to continue functioning even if a subset of RHDG cluster nodes are temporarily down.

We kindly request the development team to investigate this behavioral difference. If possible, please consider improving the recommended configuration method so that web applications can start using available nodes in the RHDG cluster, similar to the legacy method.