[
https://issues.apache.org/jira/browse/KAFKA-1112?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13825487#comment-13825487
]
Jun Rao commented on KAFKA-1112:
--------------------------------
Thinking about this a bit more. The end state is that we want to only recover
the portion of the log segment from the recovery point, instead of recovering
the whole log segment. The dilemma is that we are not sure what portion of the
index is valid. Scanning from the beginning of the log segment defeats the
purpose of incremental recovery. One possible solution is to checkpoint an
index recovery point, in addition to the recovery offset per log. The index
recovery point is the # of valid index entries in the segment to which the
recovery offset belongs. This way, on startup, we will be sure that the data in
the last valid index entry is not corrupted and we can use it to quickly locate
the recovery offset in the log file.
> broker can not start itself after kafka is killed with -9
> ---------------------------------------------------------
>
> Key: KAFKA-1112
> URL: https://issues.apache.org/jira/browse/KAFKA-1112
> Project: Kafka
> Issue Type: Bug
> Components: log
> Affects Versions: 0.8, 0.8.1
> Reporter: Kane Kim
> Assignee: Jay Kreps
> Priority: Critical
> Attachments: KAFKA-1112-v1.patch, KAFKA-1112-v2.patch,
> KAFKA-1112-v3.patch, KAFKA-1112.out
>
>
> When I kill kafka with -9, broker cannot start itself because of corrupted
> index logs. I think kafka should try to delete/rebuild indexes itself without
> manual intervention.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
