Resolution:
When  a process reports "failure detected" it means that the process has gotten a read_timeout, has issued a keep_alive, and has not received the keep_alive reply in keep_alive_timeout amount of time. The reporting client then invokes its error callback(s) to clean-up the socket and in the case of a client, will initiate a reconnect attempt to RTserver. If the RTserver is reporting this then RTserver makes no attempt to reconnect to the lost RTclient, instead the RTclient will eventually realize that it has lost its connection (it would report an error 10, or 'end of file') and will then clean up the sockect on its side and try to reconnect. In the case of RTclient it is also possible (though extremely rare) to get a 'failure detected'  if the RTclient "times out" while trying to flush a message to RTserver.

If you are seeing this error reported from the RTserver side, then the question becomes, what are the RTserver option settings for the client_read_timeout and client_keep_alive_timeout in your rtserver.cm file. A read_timeout means that the RTserver has gone that amount of time without having received any messages from a connected RTclient and that's when the RTserver sends a keep_alive mesage to RTclient,  if RTclient does not respond back in client_keep_alive_timeout seconds then RTserver reports the "failure detected"

What does the error message "Failure in reading from SPC 'protocol:_node:port:instance' {failure detected}" mean?