Following a slightly messy master election I saw a 7.6.1 cluster with the two unelected master nodes both reporting this message every 30 seconds apparently indefinitely:

failed to reassign persistent tasks org.elasticsearch.cluster.NotMasterException: no longer master. source: [reassign persistent tasks] 

Possibly there's a race in the PersistentTasksClusterService around election time? Not sure, I am not very familiar with this area.