hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ming Ma (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HDFS-10629) Federation Router
Date Thu, 29 Sep 2016 02:07:20 GMT

    [ https://issues.apache.org/jira/browse/HDFS-10629?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15531543#comment-15531543
] 

Ming Ma commented on HDFS-10629:
--------------------------------

Sorry for [~elgoiri] and [~jakace] for the late reply. Yes let us take out JMX from this patch.
Here are some comments.

* FederationNamenodeServiceState#EXPIRED isn't used. Is it needed?
* Is FederationNamenodeServiceState really needed, e.g. if HAServiceState can work?
* Do we need NamenodeStatusReport? Per https://issues.apache.org/jira/browse/HDFS-10467?focusedCommentId=15382535&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-15382535,
this might be required only for the rebalancer scenario. We can add it later when necessary.
* Router#getLocationsForPath command "Get the NN RPC client”, does it actually return RPC
client? It seems to return the remote location.
* It seems better exception handling will be handled by other jiras. For example, RetriableException
can be thrown by active NN and should be handled properly.
* Router.java has several mismatch between parameter name in the javadoc and actual name.
* ConnectionPool has clean up task to reduce the connections. But If there are lots of users,
there will be lots of ConnectionPool objects. Maybe not a major issue, but wonder if someone
can launch attack on router. Or we can make the connection max size global across different
pools. Have you checked out if Server#ConnectionManager can be used instead?
* Is DFSUtil#getNamenodeWebAddr required for this jira? If yes, support for https is needed.
* What is the use case of blockpool based lookup? For example updateBlockForPipeline and some
other methods calls invokeMethod based on blockpool. Wonder if it can use the version used
by many other methods.
* reportBadBlocks mentions DatanodeProtocol in the comment. That can be removed given Router
only serves as proxy between client and NN.
* Given ActiveNamenodeLocator and other interfaces are likely to change, maybe mark it as
@InterfaceStability.Evolving, also given they aren’t used by applications, @InterfaceAudience.Private
seems more appropriate.

> Federation Router
> -----------------
>
>                 Key: HDFS-10629
>                 URL: https://issues.apache.org/jira/browse/HDFS-10629
>             Project: Hadoop HDFS
>          Issue Type: Sub-task
>          Components: hdfs
>            Reporter: Inigo Goiri
>            Assignee: Jason Kace
>         Attachments: HDFS-10629-HDFS-10467-002.patch, HDFS-10629-HDFS-10467-003.patch,
HDFS-10629.000.patch, HDFS-10629.001.patch
>
>
> Component that routes calls from the clients to the right Namespace. It implements {{ClientProtocol}}.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-help@hadoop.apache.org


Mime
View raw message