[
https://issues.apache.org/jira/browse/HDFS-1764?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Hairong Kuang updated HDFS-1764:
--------------------------------
Attachment: deadnodesUI.patch
This is the patch for review.
1. Move startTime from DecommingStatus to DatanodeDescriptor so startTime could
be used to represent either decommissioning or dead start time;
2. Record start time when a node is declared as dead;
3. Show time since declared as dead on the deadnode web UI.
> Add 'Time Since Delared Dead' to namenode dead datanodes web page
> -----------------------------------------------------------------
>
> Key: HDFS-1764
> URL: https://issues.apache.org/jira/browse/HDFS-1764
> Project: Hadoop HDFS
> Issue Type: Improvement
> Reporter: Hairong Kuang
> Assignee: Hairong Kuang
> Attachments: deadnodesUI.patch
>
>
> I am filing this jira for Andrew. :)
> Currently on the dead nodes page of a namenode, we only list the dead
> datanode's hostnames. In addition I would like to list the duration since the
> node was declared as dead, for example in the same format as the
> "Decommissioning Nodes" page when it lists "Time Since Decommissioning
> Started".
> In our Hadoop clusters if a node has only been dead for a few minutes, our
> monitoring is likely to bring the node back without us needing to do anything
> about it. This proposed functionality will help administrators identify which
> nodes need manual attention and which nodes are likely to be fixed by our
> monitoring. If the node has been dead for many hours, it merits a closer look.
> This seems like useful functionality for open source as well.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
