C: write-transactions from all nodes cause OutOfMemoryError

The job is here:
https://jenkins.opendaylight.org/sandbox/job/controller-csit-3node-clustering-only-carbon-2nd/19/

OutOfMemoryError on node 2 can be found here:
https://logs.opendaylight.org/sandbox/jenkins091/controller-csit-3node-clustering-only-carbon-2nd/19/archives/odl2_karaf_console.log.gz

Heap dump file created [1584803866 bytes in 11.824 secs]
SLF4J: Failed toString() invocation on an object of type [org.opendaylight.controller.cluster.access.client.ConnectedClientConnection]
java.lang.OutOfMemoryError: Java heap space

The test is very simple. On all 3 nodes a write-trancactions rpc is started.
https://logs.opendaylight.org/sandbox/jenkins091/controller-csit-3node-clustering-only-carbon-2nd/19/archives/log.html.gz#s1-t1-k4

After that the robot just waits for rpc to finish. Then after 65s the rest is timed out from robot side, none of 3 responses were delivered within this timeframe.

Similar bugs:
OutOfMemoryError - https://bugs.opendaylight.org/show_bug.cgi?id=8419
not returned rpc - https://bugs.opendaylight.org/show_bug.cgi?id=8372