[
https://issues.apache.org/jira/browse/HDFS-15509?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17171551#comment-17171551
]
Ayush Saxena commented on HDFS-15509:
-------------------------------------
Hey [~LeonG],
I too have just followed that discussion, the reason there to keep both
Namenodes in same state is that when an admin triggers the {{safemode}}
command, he expects that now the cluster is in readonly mode, the cluster won't
respond to any write calls, but in case, if one namenode is down and ignored,
In that case, If the other Namenode comes alive, after this safemode command
execution, and becomes active namenode due to failover, then the objective to
make the cluster read-only won't hold and the cluster shall start serving the
write request as well.
This is one perspective, there are indeed many ways of looking at it, if you
tend to have some opinions feel free to share them at HDFS-8277
> Set safemode should not fail if one of the namenode is down.
> ------------------------------------------------------------
>
> Key: HDFS-15509
> URL: https://issues.apache.org/jira/browse/HDFS-15509
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: hdfs
> Affects Versions: 3.3.0
> Reporter: Leon Gao
> Assignee: Leon Gao
> Priority: Minor
> Attachments: HDFS-15509.patch
>
>
> When the first namenode (let's say nn0) is down, set safemode command will
> always fail unless users manually update the configuration. This is
> distracting when debugging issues.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
