currently the schema for the clustering mode says it's a string and
the mode is decided by first letter of the string. this means that any garbage starting with the correct letter may lead to use of a specific mode e.g.:

lasjadgadfg -> local
rhdbhgh -> replication
dfghbmjfgh -> distribution
IlkHhdf -> invalidation

what about forcing only the documented values
"For distribution, set mode to either 'd', 'dist' or 'distribution'. For replication, use either 'r', 'repl' or 'replication'. Finally, for invalidation, 'i', 'inv' or 'invalidation'."
+ LOCAL