[
https://issues.apache.org/jira/browse/HDFS-6329?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Kihwal Lee updated HDFS-6329:
-----------------------------
Attachment: HDFS-6329.patch
The idea was sound, but didn't work with the way MiniDFSCluster is setting up
namenode. I modified the patch to
* have MiniDFSCluster set fs.defaultFs to "hdfs://127.0.0.1:0" instead of
"127.0.0.1:0".
* make NameNode delay the setting of clientNamenodeAddress, if it is binding to
a random port.
> WebHdfs does not work if HA is enabled on NN but logical URI is not
> configured.
> -------------------------------------------------------------------------------
>
> Key: HDFS-6329
> URL: https://issues.apache.org/jira/browse/HDFS-6329
> Project: Hadoop HDFS
> Issue Type: Bug
> Affects Versions: 2.4.0
> Reporter: Kihwal Lee
> Assignee: Kihwal Lee
> Priority: Blocker
> Attachments: HDFS-6329.patch, HDFS-6329.patch
>
>
> After HDFS-6100, namenode unconditionally puts the logical name (name service
> id) as the token service when redirecting webhdfs requests to datanodes, if
> it detects HA.
> For HA configurations with no client-side failover proxy provider (e.g. IP
> failover), webhdfs does not work since the clients do not use logical name.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
