[
https://issues.apache.org/jira/browse/HDFS-9133?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14908720#comment-14908720
]
Colin Patrick McCabe commented on HDFS-9133:
--------------------------------------------
bq. Thanks Colin for the work. The patch itself looks good, +1.
Thanks
bq. For other block readers, the length means the number of bytes to read, but
in ExternalBlockReader, it means "visible length" which assumes it's the block
replica length? Since from the calculation of skip and available in
ExternalBlockReader , it indicates it assumes it's the block replica length.
You are correct... we should not be treating this as the {{visibleLength}} of
the block, since it's really the number of bytes in between the end of the
block and the current offset within the block. Let me file another jira to fix
this.
> ExternalBlockReader and ReplicaAccessor need to return -1 on read when at EOF
> -----------------------------------------------------------------------------
>
> Key: HDFS-9133
> URL: https://issues.apache.org/jira/browse/HDFS-9133
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: hdfs-client
> Affects Versions: 2.8.0
> Reporter: Colin Patrick McCabe
> Assignee: Colin Patrick McCabe
> Attachments: HDFS-9133.001.patch
>
>
> ExternalBlockReader and ReplicaAccessor need to return -1 on read when at
> EOF, as per the JavaDoc in BlockReader.java.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
