[
https://issues.apache.org/jira/browse/KAFKA-1112?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13814134#comment-13814134
]
Guozhang Wang commented on KAFKA-1112:
--------------------------------------
Attached another stack trace with more debugging turned on. The root cause is
that when we load the index file, the initial size is set to the limit of the
file, and hence the position is pointing to the last entry. In most cases the
last entry will be 0, and the recoveryLog process will skip since it shows the
latest offset is smaller than the checkpoint. But when doing the sanity check
it finds the number of entries is non-zero (actually the max number of entries)
while the last offset is equal to the base offset since reading the last entry
gives you 0 relative offset.
> 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
>
> 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)
