From Drools-5.0.0, Expert, 7.8.1. Sentence numbers added by me:

[1] when a ruleflow-group becomes active or an agenda-group receives the focus
any rules that have lock-on-active set to try cannot place activations onto the
agenda, the rules are matched and the resulting activations discarded.
[2] This is a stronger version of no-loop.
[3] It's ideally for calculation rules where you have a number of rules that
will modify a fact and you don't want any rule re-matching and firing.
[4] In summary fire these currently active rules and only these rules, no
matter how the data changes, do not allow any more activations for the rules
with the attribute set to true.
[5] When the ruleflow-group is no longer active or agenda-group loses the
focus those rules with lock-on-active set to true can once again add
activations onto the agenda.

[4] says that activations "for the rules with the attribute set to true" are
disallowed. This is confusing or even in contradiction with [1], which I
understand to mean that WM updates done by the RHS of a rule X with
lock-on-active==true don't result in any activations. (Of course, this
includes activations of X itself.)

[5] is absolutely confusing: How can rules in a group that isn't active
or doesn't have the focus can add activations - irrespective of its
lock-on-active setting?

[1] s/when/When/, s/try/true/

I would create a JIRA, but I'd like to be sure what to suggest for a fix.