[
https://issues.apache.org/jira/browse/KAFKA-1028?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13881297#comment-13881297
]
Jason Rosenberg commented on KAFKA-1028:
----------------------------------------
Thanks for the explanation [~noslowerdna]. So, this will need to be a property
for the broker, and not the producer config, I see (so setting
request.required.acks on the producer won't help with deciding to do unclean
leader election).
So, my next question, with this option, will we still be able to use whatever
setting for request.required.acks from the producer? Seems like the answer is
yes (and it doesn't change the case if you set acks to 0, and the topic is
currently not available, then you wouldn't actually know if the message got
published, which is the case anyway).
I suppose another trade-off too, is what happens if the A and B not only go
down (in your example above), but end up in an unrecoverable state? E.g. they
caught on fire, the disks are toast. Will the topic ever be able to recover,
in this case?
> per topic configuration of preference for consistency over availability
> -----------------------------------------------------------------------
>
> Key: KAFKA-1028
> URL: https://issues.apache.org/jira/browse/KAFKA-1028
> Project: Kafka
> Issue Type: Improvement
> Reporter: Scott Clasen
> Assignee: Neha Narkhede
>
> As discussed with Neha on the ML.
> It should be possible to configure a topic to disallow unclean leader
> election, thus preventing the situation where committed messages can be
> discarded once a failed leader comes back online in a situation where it was
> the only ISR.
> This would open kafka to additional usecases where the possibility of
> committted messages being discarded is unacceptable.
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
