Fabric Ensemble Server fails to rejoin the Fabric after a container restart

Issue

Note:
This KCS article documents a very specific Fabric related problem.
There can be other reasons for the high level problem of a Fabric managed container not joining the Fabric upon restart or upon some other activity. Please read all sections of this KCS carefully to determine if you suffer the same problem.

A Fabric ensemble server had to be restarted but now fails to rejoin the Fabric.
While the container starts up, it repeatedly raises errors like the following

2019-04-29 12:49:42,674 | ERROR | c]-nio2-thread-2 | fabric-core                      | 70 - io.fabric8.fabric-core - 1.2.0.redhat-630371 | [io.fabric8.service(37)] The activate method has thrown an exception
java.lang.NullPointerException
[... stack trace omitted ...]
2019-04-29 12:49:42,676 | ERROR | c]-nio2-thread-2 | fabric-core                      | 70 - io.fabric8.fabric-core - 1.2.0.redhat-630371 | [io.fabric8.service(37)] Failed creating the component instance; see log for reason
2019-04-29 12:49:42,678 | ERROR | c]-nio2-thread-2 | fabric-core                      | 70 - io.fabric8.fabric-core - 1.2.0.redhat-630371 | [io.fabric8.mbeanserver.listener(15)] Cannot create component instance due to failure to bind reference runtimeProperties
2019-04-29 12:49:42,678 | ERROR | c]-nio2-thread-2 | fabric-core                      | 70 - io.fabric8.fabric-core - 1.2.0.redhat-630371 | [io.fabric8.mbeanserver.listener(15)] Failed creating the component instance; see log for reason

See the Diagnostics section for a full stack trace.

When running fabric:container-list from another Fabric ensemble server, this container shows as not running and not connected.
Manually restarting this container does not help to resolve the problem.

The problem affects only one out of three or more ensemble servers. The other ensemble server still form a quorum and as a result the Fabric registry is still working.