Incorrect behavior Oozie with increased load. Hanging subsidiaries in the status "RUNNING"

Question

I launch an Oozie workflow with the following structure:

-- Oozie workflow
------> subworkflow_1
---------- fork_1
---------- fork_2
---------- ...
---------- fork_n
------> subworkflow_2
---------- fork_1
---------- fork_2
---------- ...
---------- fork_n

And with a high load on the cluster, I come across a problem. one of the fork have status "RUNNING" but if you open this fork, then it has "SUCCESS" status. I get such an error even eat in every fork I read the Spark Pi.

I do not understand why the subsidiary does not convey to the parental status on time. Tell me, are there any parameters for Oozie, with which I can regulate the interval of a survey of subsidiaries to find out their status?

Parameter oozie.validate.ForkJoin = false does not solve my problem.