[
https://issues.apache.org/jira/browse/KAFKA-2188?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jun Rao updated KAFKA-2188:
---------------------------
Status: In Progress (was: Patch Available)
[~tnachen], thanks for the patch.
It seems that KIP-18 is still under discussion and hasn't been approved. A
couple of thoughts on this.
1. Instead of auto re-replicating the partitions on the failed disk to some
other disks, I was thinking that a less aggressive approach is to simply fail
those affected replicas w/o triggering the re-replication, but keep the broker
running. The controller will still be informed of those failed replicas, but
won't restart them. If those failed replicas are leaders, the controller will
move the leaders to other replicas. This approach is probably simpler and gives
us the main benefit, which is you don't hard-fail a broker immediately on
single disk failure. The admin can cleanly shut down this broker at the
appropriate time and fix the disk issue.
2. For 0.8.3 release, we are mainly trying to get the new consumer and some of
the security features done. Could we defer this until 0.8.3 is out?
> JBOD Support
> ------------
>
> Key: KAFKA-2188
> URL: https://issues.apache.org/jira/browse/KAFKA-2188
> Project: Kafka
> Issue Type: Bug
> Reporter: Andrii Biletskyi
> Assignee: Andrii Biletskyi
> Attachments: KAFKA-2188.patch, KAFKA-2188.patch, KAFKA-2188.patch
>
>
> https://cwiki.apache.org/confluence/display/KAFKA/KIP-18+-+JBOD+Support
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
