[ https://issues.apache.org/jira/browse/KAFKA-7996?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16778486#comment-16778486 ]
Patrik Kleindl commented on KAFKA-7996: --------------------------------------- [~mjsax] [~guozhang] Apologies for creating this stir, you are right that KafkaStreams.close is behaving as it should. I got too focused on the producer timeout and mixed things up. It seems we (and our code) take the "false" from the close method too strict as we interpreted this as a kind of fatal condition and don't try to start a new instance of the streams application as we can't be sure if the shutdown was successful. We are not running in the cloud so a proper shutdown seemed important. As it stands even with proper timeouts for the producer and adminClient the close can take up to several minutes in the worst case which is fine. So my hopefully last questions: If a KafkaStreams instance is in ERROR state, is it safe to start a new one or should one wait until the close() is completed? Can the close() change anything regarding topics or state stores which might prevent a new instance from starting or would it be better to call .cleanup() on the new instance? > KafkaStreams does not pass timeout when closing Producer > -------------------------------------------------------- > > Key: KAFKA-7996 > URL: https://issues.apache.org/jira/browse/KAFKA-7996 > Project: Kafka > Issue Type: Bug > Components: streams > Affects Versions: 2.1.0 > Reporter: Patrik Kleindl > Assignee: Lee Dongjin > Priority: Major > Labels: needs-kip > > [https://confluentcommunity.slack.com/messages/C48AHTCUQ/convo/C48AHTCUQ-1550831721.026100/] > We are running 2.1 and have a case where the shutdown of a streams > application takes several minutes > I noticed that although we call streams.close with a timeout of 30 seconds > the log says > [Producer > clientId=…-8be49feb-8a2e-4088-bdd7-3c197f6107bb-StreamThread-1-producer] > Closing the Kafka producer with timeoutMillis = 9223372036854775807 ms. > Matthias J Sax [vor 3 Tagen] > I just checked the code, and yes, we don't provide a timeout for the producer > on close()... -- This message was sent by Atlassian JIRA (v7.6.3#76005)