[
https://issues.apache.org/jira/browse/KAFKA-1296?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Neha Narkhede updated KAFKA-1296:
---------------------------------
Description: In the new mirror maker (like the old MM), the MM
asynchronously sends data through the producer. Due to this, it doesn't have a
way to get any exception that the producer sends back and do something
meaningful. At least not without falling back to synchronous behavior. What
this means is that if the producer runs out of retries, it silently drops data
on the floor and the user is left wondering why the # of messages in the target
cluster don't match with those in the source cluster. (was: In the new mirror
maker (like the old MM), the MM asynchronously sends data through the producer.
Due to this, it doesn't have a way to get any exception that the producer sends
back and do something meaningful. At least not without falling back to
synchronous behavior. What this means is that if the producer runs out of
retries, it silently drops data on the floor and the user is left wondering why
the # of messages in the target cluster don't match with those in the source
cluster.
I think it is useful for the producer to log an error before returning the
exception to the user. )
> Improve error reporting in new mirror maker
> -------------------------------------------
>
> Key: KAFKA-1296
> URL: https://issues.apache.org/jira/browse/KAFKA-1296
> Project: Kafka
> Issue Type: Improvement
> Components: producer
> Affects Versions: 0.9.0
> Reporter: Neha Narkhede
> Assignee: Neha Narkhede
> Priority: Critical
>
> In the new mirror maker (like the old MM), the MM asynchronously sends data
> through the producer. Due to this, it doesn't have a way to get any exception
> that the producer sends back and do something meaningful. At least not
> without falling back to synchronous behavior. What this means is that if the
> producer runs out of retries, it silently drops data on the floor and the
> user is left wondering why the # of messages in the target cluster don't
> match with those in the source cluster.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
