[
https://issues.apache.org/jira/browse/HDFS-1623?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13070622#comment-13070622
]
Suresh Srinivas commented on HDFS-1623:
---------------------------------------
The problem is not with the timing of delivery of disconnect event. During fail
over, the standby taking over as active may not be able to communicate with
previous active (directly/indirectly) to assert that the previous active has
relinquished the role of active. This could be due to network partition, active
not functional due to GC, OS issues etc. In such a scenario, the only way for
new active to ensure shared resource is not controlled by two actives is to
fence the shared resource.
> High Availability Framework for HDFS NN
> ---------------------------------------
>
> Key: HDFS-1623
> URL: https://issues.apache.org/jira/browse/HDFS-1623
> Project: Hadoop HDFS
> Issue Type: New Feature
> Reporter: Sanjay Radia
> Assignee: Sanjay Radia
> Attachments: HDFS-High-Availability.pdf, NameNode HA_v2.pdf, NameNode
> HA_v2_1.pdf, Namenode HA Framework.pdf
>
>
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
