This is a follow-up to the discussion in JBEAP-29870, to track the improvement of JBoss EAP 8.1.0 on OpenShift documentation, specifically about the different cluster behavior upon a node crash or removal, and how it is detected.

The current documentation could be misleading for users, and it should be improved by explicitly mentioning differences between service failures and pod deletions and the way failure detection works in each of the aforementioned cases (i.e. JGroups FD_SOCK protocol for crash, FD_ALL3 for pod deletion).

A note could cover StatefulSet based clusters specifically, where pod deletion is described by the Kubernetes documentation as an event which has unpredictable consequences with respect to data loss.

original link to JBEAP
https://issues.redhat.com/browse/JBEAP-30590