Shashikant Banerjee commented on HDDS-161:

As per the offline discussion with [~anu], storing the replication type info in 
the containerdata in the datanodes is not required. In patch v1, while sending 
the CloseCommand in the hearbeart response to datanode, the replication type 
info is also sent so that, the closeContainer command can be queued to the 
appropriate Xeciver server on the Datanode to handle it.

> Add functionality to queue ContainerClose command from SCM Hearbeat Reposnse 
> to Ratis
> -------------------------------------------------------------------------------------
>                 Key: HDDS-161
>                 URL: https://issues.apache.org/jira/browse/HDDS-161
>             Project: Hadoop Distributed Data Store
>          Issue Type: Bug
>          Components: Ozone Datanode, SCM
>            Reporter: Shashikant Banerjee
>            Assignee: Shashikant Banerjee
>            Priority: Major
>             Fix For: 0.2.1
>         Attachments: HDDS-161.00.patch, HDDS-161.01.patch
> When a container needs to be closed at the Datanode, SCM will queue a close 
> command which will be encoded as a part of Heartbeat Response to the 
> Datanode. This command will be picked up from the response at the Datanode 
> which will then be submitted to the XceiverServer to process the close 
> command. This will just queue a ContainerCloseCommand to the Ratis, where the 
> leader would start the transaction while the followers will reject the 
> closeContainer request.
> While handling the close container inside the Datanode, we need to ensure all 
> the ongoing chunkWrites finish before close can proceed through. It should 
> also reject any any incoming I/Os in between. This will be handled as a part 
> of separate jira.

This message was sent by Atlassian JIRA

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