You are viewing a plain text version of this content. The canonical link for it is here.

Posted to hdfs-dev@hadoop.apache.org by "Hairong Kuang (JIRA)" <ji...@apache.org> on 2011/01/11 01:46:45 UTC

[jira] Created: (HDFS-1577) Fall back to a random datanode when bestNode fails

Fall back to a random datanode when bestNode fails
--------------------------------------------------

                 Key: HDFS-1577
                 URL: https://issues.apache.org/jira/browse/HDFS-1577
             Project: Hadoop HDFS
          Issue Type: Bug
          Components: name-node
            Reporter: Hairong Kuang
            Assignee: Hairong Kuang
             Fix For: 0.23.0


When NameNod decides to redirect a read request to a datanode, if it can not find a live node that contains a block of the file, NameNode should choose a random datanode instead of throwing an exception.

This is because a live node test is against its http port. A non-functional jetty servlet (may due to bug like JETTY-1264) does not mean that the replica on that DataNode is not readable. Redirecting the read request to a random datanode could make hftp function better when DataNodes hit bugs like JETTY-1264.

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