[
https://issues.apache.org/jira/browse/HDFS-14998?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16984031#comment-16984031
]
Chao Sun commented on HDFS-14998:
---------------------------------
Good point. I think we are on the same page then. I have a few comments on the
latest patch:
1.
{quote}
In general ZKFC shouldn't be turned on the Observer NameNode since it doesn't
add any such value, though now it is not mandatory to do so
{quote}
How about:
{quote}
ZKFC does not need to be turned on the Observer NameNode unless transition
between between Observer and Standby is required in your cluster
{quote}
2.
{quote}
If **dfs.ha.automatic-failover.enabled** is turned on, you could run ZKFC on
the namenode for observer, but it is not recommended because the Observer
NameNode will not participate in failover. In addition to that, you'll also
need to add **forcemanual** flag to the **transitionToObserver** command:
{quote}
How about:
{quote}
If **dfs.ha.automatic-failover.enabled** is turned on, the only benefit for
running ZKFC on Observer NameNode is to enable dynamic transition from Observer
to Standby role (which will then automatically join the Zookeeper controlled
failover group) and vise versa. If this is not desired, you can disable ZKFC on
the Observer NameNode.
In addition, currently a **forcemanual** flag is required with the
**transitionToObserver** command:
{quote}
3.
{quote}
After the namenode is transitioned to observer state, you could run ZKFC on
this namenode if you want, but ZKFC doesn't do anything right now. When you
transition the namenode to standby state, ZKFC running on this namenode will
participate in automatic failover.
{quote}
This paragraph may not be necessary with the above change.
Let me know what you think. Thanks.
> Update Observer Namenode doc for ZKFC after HDFS-14130
> ------------------------------------------------------
>
> Key: HDFS-14998
> URL: https://issues.apache.org/jira/browse/HDFS-14998
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: documentation
> Affects Versions: 3.3.0
> Reporter: Fei Hui
> Assignee: Fei Hui
> Priority: Minor
> Attachments: HDFS-14998.001.patch, HDFS-14998.002.patch,
> HDFS-14998.003.patch
>
>
> After HDFS-14130, we should update observer namenode doc, observer namenode
> can run with ZKFC running
--
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
