[
https://issues.apache.org/jira/browse/FLUME-2117?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13844467#comment-13844467
]
Hari Shreedharan commented on FLUME-2117:
-----------------------------------------
If the error is deterministically reproducible (maybe bad event, or serializer
throws always, this needs to be handled by the serializer code). FLUME-2259
takes care of a bug in the sink that can cause the sink to die and never
actually process an event once an exception is thrown. Once that is committed,
that is issue would go away
> HBaseSink & flume.SinkRunner - Infinite loop
> ---------------------------------------------
>
> Key: FLUME-2117
> URL: https://issues.apache.org/jira/browse/FLUME-2117
> Project: Flume
> Issue Type: Bug
> Components: Sinks+Sources
> Affects Versions: v1.3.0
> Reporter: Charles Orth
>
> HBaseSink.process method calls action to pull an action from the channel. If
> our sink throws an exception, the transaction is never closed and the current
> action remains on the channel. Thereby, repeatedly replaying the same action.
> The flume process must have manual intervention to remove the invalid action.
--
This message was sent by Atlassian JIRA
(v6.1.4#6159)