This log message ...
... could use some improvement.
First of all, it is logged per-JRuby, and that's confusing - it looks like the cache is being invalidated N times (where N is the number of JRubies); and technically, it is, since the caching is per-JRuby; but if you just see a block of those messages go by, it can be confusing - especially so for an end-user who might not understand the guts of Puppet Server.
So, perhaps it should only be logged once at INFO level for the entire operation, and the per-JRuby messages could be bumped down to DEBUG. It would also be clearer if they mentioned that they were per-JRuby, perhaps by including the JRuby instance ID.