[ 
https://issues.apache.org/jira/browse/HDDS-348?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16596045#comment-16596045
 ] 

Shashikant Banerjee commented on HDDS-348:
------------------------------------------

Thanks [~msingh], for reporting and working on this issue. The patch overall 
looks good to me.
 # When stateMachineHelper is instantiated, we can assign the executor to it 
right away as it only handles operations for a particular container. That way , 
it will not be required to find which executor it needs to execute the commands 
in during execution.
 # We can remove writeChunkExecutor and use the container specific executor 
added with the patch to perform write chunk operations . 
 # XceiverServerRatis.java : L140-144 : Instead of of defining a singkle 
threaded executor instance we can have a configurable no of threads per 
executor. Default value may be set to 1.
 # ScmConfigKeys.java: 60 : "dfs.container.ratis.num.container.op.threads" is 
actually the no of executors being used rather than no of threads. The config 
name should be changed accordingly. Same thing applies to changes in 
OzoneConfigKeys.java.
 # It would be good to add some more comments in ContainerStateMachine class 
regarding work flow with executors.

> Parallalize container ops in ContainerStateMachine
> --------------------------------------------------
>
>                 Key: HDDS-348
>                 URL: https://issues.apache.org/jira/browse/HDDS-348
>             Project: Hadoop Distributed Data Store
>          Issue Type: Bug
>          Components: Ozone Datanode
>    Affects Versions: 0.2.1
>            Reporter: Mukul Kumar Singh
>            Assignee: Mukul Kumar Singh
>            Priority: Major
>             Fix For: 0.2.1
>
>         Attachments: HDDS-348.001.patch, HDDS-348.002.patch
>
>
> Currently all the ops in ContainerStateMachine#applyTransaction are processed 
> sequentially by the ContainerStateMachine. However these ops can be 
> parallelized by having a configurable no of executors where a certain 
> executor for a container can be chosen from the pool of executors by hashing 
> the containerID.



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

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to