[
https://issues.apache.org/jira/browse/HDFS-9574?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Kihwal Lee updated HDFS-9574:
-----------------------------
Attachment: HDFS-9574.patch
Since {{DataXceiverServer}} is shared by multiple block pools, we cannot simply
move its initialization to later point. I will have each data transfer op wait
up to configured amount of time before giving up. A number of threads will pile
up until the node registers, but it cannot serve anyway, so I don't think we
are introducing any more service degradation.
> Reduce client failures during datanode restart
> ----------------------------------------------
>
> Key: HDFS-9574
> URL: https://issues.apache.org/jira/browse/HDFS-9574
> Project: Hadoop HDFS
> Issue Type: Bug
> Reporter: Kihwal Lee
> Assignee: Kihwal Lee
> Attachments: HDFS-9574.patch
>
>
> Since DataXceiverServer is initialized before BP is fully up, client requests
> will fail until the datanode registers.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
