[
https://issues.apache.org/jira/browse/HDFS-14427?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16834412#comment-16834412
]
Fengnan Li commented on HDFS-14427:
-----------------------------------
[~elgoiri] I gave this a second thought about the idea of specifying the number
of Routers, but didn't find much point of doing this. See the total namenodes
from all nameservices is X, if we set the number < X, we will need to manually
remove some routers from the RouterContext collection; if we set the number >
X, we have to replicate some RouterContext. From the code level all of this are
doable, but from the actual usage, it seems not very meaningful.
The uploaded patch is only for comment change.
> RBF: Optimize some testing set up logic in MiniRouterDFSCluster
> ---------------------------------------------------------------
>
> Key: HDFS-14427
> URL: https://issues.apache.org/jira/browse/HDFS-14427
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Reporter: Fengnan Li
> Assignee: Fengnan Li
> Priority: Major
> Attachments: HDFS-14427-HDFS-13891.001.patch
>
>
> [https://github.com/apache/hadoop/blob/HDFS-13891/hadoop-hdfs-project/hadoop-hdfs-rbf/src/test/java/org/apache/hadoop/hdfs/server/federation/MiniRouterDFSCluster.java#L808]
> the comment says one router is created per name service, while in the code
> one router is created per namenode in each nameservice.
> There are a couple of things that might need to consider optimization:
> # make the code as the the comment
> # add some ways to specify the number of routers
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
