hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ajay Kumar (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HDDS-228) Add the ReplicaMaps to ContainerStateManager
Date Tue, 10 Jul 2018 19:44:00 GMT

    [ https://issues.apache.org/jira/browse/HDDS-228?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16539160#comment-16539160
] 

Ajay Kumar commented on HDDS-228:
---------------------------------

[~nandakumar131] thanks for review. Addressed most of you suggestions, except one.
{quote}ContainerStateManager#getContainerReplicas javadoc is incorrect.
ContainerStateMap#getContainerReplicas javadoc is incorrect
ContainerStateMap#removeContainerReplica: 
For the third case, the call should throw Exception.{quote}
Done
{quote}ContainerStateMap#addContainerReplica: If the container is not already present, this
method adds the container to the contReplicaMap and updates the set with the given datanode.
It would be easy to detect inconsistency if we throw Exception in case if the container is
not present in the contReplicaMap. A container should be added to the contReplicaMap when
addContainer is called. This way we can maintain consistency between containerMap and contReplicaMap.{quote}
IMO adding a new replica even if one doesn't exist do not violate any condition. So we should
add it silently. Also there is little value in adding open containers to replicaMap as there
replication will be handled by RATIS. So only contrainer report handler should add or remove
entries from this instead of adding it via addContainer.


> Add the ReplicaMaps to ContainerStateManager
> --------------------------------------------
>
>                 Key: HDDS-228
>                 URL: https://issues.apache.org/jira/browse/HDDS-228
>             Project: Hadoop Distributed Data Store
>          Issue Type: Improvement
>          Components: SCM
>            Reporter: Anu Engineer
>            Assignee: Ajay Kumar
>            Priority: Major
>             Fix For: 0.2.1
>
>         Attachments: HDDS-228.00.patch, HDDS-228.01.patch, HDDS-228.02.patch, HDDS-228.03.patch,
HDDS-228.04.patch
>
>
> We need to maintain a list of data nodes in the SCM that tells us where a container is
located. This created from the container reports.  The HDDS-175 refactored the class to make
this separation easy and this JIRA is a followup that keeps a hash table to track this information.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

---------------------------------------------------------------------
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