hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Todd Lipcon (JIRA)" <j...@apache.org>
Subject [jira] Updated: (HDFS-378) DFSClient should track failures by datanode across all streams
Date Thu, 01 Apr 2010 23:02:27 GMT

     [ https://issues.apache.org/jira/browse/HDFS-378?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Todd Lipcon updated HDFS-378:
-----------------------------

    Description: 
Original description: Rather than tracking the total number of times DFSInputStream failed
to talk to a datanode for a particular block, such failures and the the list of datanodes
involved should be scoped to individual blocks. In particular, the "deadnode" list should
be a map of blocks to a list of failed nodes, the latter reset and the nodes retried per the
existing semantics.

[see comment below for new thinking, left this comment to give context to discussion]

  was:Rather than tracking the total number of times DFSInputStream failed to locate a datanode
for a particular block, such failures and the the list of datanodes involved should be scoped
to individual blocks. In particular, the "deadnode" list should be a map of blocks to a list
of failed nodes, the latter reset and the nodes retried per the existing semantics.

        Summary: DFSClient should track failures by datanode across all streams  (was: DFSClient
should track failures by block rather than globally)

> DFSClient should track failures by datanode across all streams
> --------------------------------------------------------------
>
>                 Key: HDFS-378
>                 URL: https://issues.apache.org/jira/browse/HDFS-378
>             Project: Hadoop HDFS
>          Issue Type: Improvement
>            Reporter: Chris Douglas
>
> Original description: Rather than tracking the total number of times DFSInputStream failed
to talk to a datanode for a particular block, such failures and the the list of datanodes
involved should be scoped to individual blocks. In particular, the "deadnode" list should
be a map of blocks to a list of failed nodes, the latter reset and the nodes retried per the
existing semantics.
> [see comment below for new thinking, left this comment to give context to discussion]

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message