[
https://issues.apache.org/jira/browse/KAFKA-2998?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15093078#comment-15093078
]
Jason Gustafson commented on KAFKA-2998:
----------------------------------------
[~ijuma] Maybe we can treat this as a separate issue. It seems common to
accidentally point the bootstrap brokers to zookeeper, for example, and the
current behavior is to silently retry forever. Maybe we can address the common
misconfiguration problem here without needing the full patch for KAFKA-1894? A
simple approach might be to log disconnects to any of the bootstrap brokers at
the error level.
> New Consumer should not retry indefinitely if no broker is available
> --------------------------------------------------------------------
>
> Key: KAFKA-2998
> URL: https://issues.apache.org/jira/browse/KAFKA-2998
> Project: Kafka
> Issue Type: Bug
> Components: clients
> Affects Versions: 0.9.0.0
> Reporter: Florian Hussonnois
> Assignee: Jason Gustafson
> Priority: Minor
>
> If no broker from bootstrap.servers is available consumer retries
> indefinitely with debug log message :
>
> DEBUG 17:16:13 Give up sending metadata request since no node is available
> DEBUG 17:16:13 Initialize connection to node -1 for sending metadata request
> DEBUG 17:16:13 Initiating connection to node -1 at localhost:9091.
> At least, an ERROR message should be log after a number of retries.
> In addition, maybe the consumer should fail in a such case ? This behavior
> could be set by a configuration property ?
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
