[ https://issues.apache.org/jira/browse/KAFKA-10303?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17169288#comment-17169288 ]
Yogesh BG commented on KAFKA-10303: ----------------------------------- Hi this issue is resolved now, we had an issue with node ip assignment, due to which we used to get this issue. Thanks. we can close this issue > kafka producer says connect failed in cluster mode > -------------------------------------------------- > > Key: KAFKA-10303 > URL: https://issues.apache.org/jira/browse/KAFKA-10303 > Project: Kafka > Issue Type: Bug > Reporter: Yogesh BG > Priority: Major > > Hi > > I am using kafka broker version 2.3.0 > We have two setups with standalone(one node) and 3 nodes cluster > we pump huge data ~25MBPS, ~80K messages per second > It all works well in one node mode > but in case of cluster, producer start throwing connect failed(librd kafka) > after sometime again able to connect start sending traffic. > What could be the issue? some of the configurations are > > replica.fetch.max.bytes=10485760 > num.network.threads=12 > num.replica.fetchers=6 > queued.max.requests=50000 > # The number of threads doing disk I/O > num.io.threads=12 > replica.socket.receive.buffer.bytes=10000 > > -- This message was sent by Atlassian Jira (v8.3.4#803005)