[ https://issues.apache.org/jira/browse/NIFI-8357?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17307843#comment-17307843 ]
ASF subversion and git services commented on NIFI-8357: ------------------------------------------------------- Commit 2f08d1f466b9f6f0b0b8a7b5893341a0d1433a4e in nifi's branch refs/heads/main from Mark Payne [ https://gitbox.apache.org/repos/asf?p=nifi.git;h=2f08d1f ] NIFI-8357: Updated Kafka 2.0 processors to automatically handle recreating Consumer Lease objects when an existing one is poisoned, even if using statically assigned partitions This closes #4926. Signed-off-by: Peter Turcsanyi <turcsa...@apache.org> > ConsumeKafka(Record)_2_0, ConsumeKafka(Record)_2_6 do not reconnect if using > statically assigned partitions > ----------------------------------------------------------------------------------------------------------- > > Key: NIFI-8357 > URL: https://issues.apache.org/jira/browse/NIFI-8357 > Project: Apache NiFi > Issue Type: Bug > Components: Extensions > Reporter: Mark Payne > Assignee: Mark Payne > Priority: Critical > Fix For: 1.14.0 > > Time Spent: 10m > Remaining Estimate: 0h > > If using statically assigned partitions in ConsumeKafka_2_0, > ConsumeKafkaRecord_2_0, ConsumeKafka_2_6, or ConsumeKafkaRecord_2_6 (via > adding {{partitions.}}{{<hostname}}> properties), when a client connection > fails, it recreates connections but does not properly assign the partitions. > As a result, the consumer stops consuming data from its partition(s), and the > Kafka client that gets created gets leaked. This can slowly build up to > leaking many of these connections and potentially could exhaust heap or cause > IOException: too many open files. -- This message was sent by Atlassian Jira (v8.3.4#803005)