In my experience, the retry loop pattern has led to a deadlock when the number of failed flow files approaches the size of the queues entering and exiting the loop. The issue is that both of those queues fill up, and NiFi is not able to pick up on the fact that it can cycle flowfiles between the queues. This also has the unfortunate side effect of blocking any flowfiles from entering the initial error-prone processor, as they could potentially fail, and enter a full queue. The best solution I have as of now is to simply duplicate any failure prone processors, considering I usually only want to retry 1-3 times. See: http://apache-nifi-users-list.2361937.n4.nabble.com/Back-pressure-deadlock-td3274.html ... View more