[ 
https://issues.apache.org/jira/browse/HDDS-733?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16675599#comment-16675599
 ] 

Nanda kumar commented on HDDS-733:
----------------------------------

[~ljain], thanks for working on this. The patch looks good to me, couple of 
minor comments

In SCMChillModeManager#ContainerChillModeRule we should exclude the containers 
which are in OPEN state from adding to containerMap. Now the containers in OPEN 
state might not be created in datanode.
ChillModeManager should track pipelines in the cluster for containers in OPEN 
state.

 

Unused imports in SCMContainerManager and TestDeadNodeHandler.

 

As [~linyiqun] pointed out, we can add a test case to make sure that we don't 
create containers for ReadChunk.
{quote}Send ReadChunk request before WriteChunk request and verify the 
StorageContainerException of CONTAINER_NOT_FOUND.
{quote}
 

Looks the following tests are failing with this patch, can you take a look?
* org.apache.hadoop.ozone.client.rpc.TestCloseContainerHandlingByClient
* org.apache.hadoop.ozone.freon.TestFreonWithDatanodeRestart

> Create container if not exist, as part of chunk write
> -----------------------------------------------------
>
>                 Key: HDDS-733
>                 URL: https://issues.apache.org/jira/browse/HDDS-733
>             Project: Hadoop Distributed Data Store
>          Issue Type: Improvement
>          Components: Ozone Datanode
>            Reporter: Nanda kumar
>            Assignee: Lokesh Jain
>            Priority: Major
>         Attachments: HDDS-733.001.patch, HDDS-733.002.patch
>
>
> The current implementation requires a container to be created in datanode 
> before starting the chunk write. This can be optimized by creating the 
> container on the first chunk write.
>  During chunk write, if the container is missing, we can go ahead and create 
> the container.
> Along with this change ALLOCATED and CREATING container states can be removed 
> as they were used to track which containers have been successfully created. 
> Also there is a shouldCreateContainer flag which is used by client to know if 
> it needs to create container. This flag can be removed.



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

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

Reply via email to