You are viewing a plain text version of this content. The canonical link for it is here.
Posted to hdfs-issues@hadoop.apache.org by "Shashikant Banerjee (JIRA)" <ji...@apache.org> on 2018/10/09 08:02:00 UTC
[jira] [Comment Edited] (HDDS-550) Serialize ApplyTransaction calls
per Container in ContainerStateMachine
[ https://issues.apache.org/jira/browse/HDDS-550?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16642881#comment-16642881 ]
Shashikant Banerjee edited comment on HDDS-550 at 10/9/18 8:01 AM:
-------------------------------------------------------------------
Thanks [~xyao], for the review comments. Patch v3 fixes the compilation failure. The idea is to add the writeChunkFuture in the futureMap and once it finishes should remove it from the Map. Patch v3 has the correct ordering.
Prior to the patch, it was possible that applyTransaction for a write chunk call will start even before the writeChunk task in #writeStateMachineData completes. Hence, writeChunkFuture was getting removed from the Map during applyTransaction. But Now, it is guaranteed to be complete before applyTransaction call comes via Ratis, that is why the idea in the current patch is to remove the writeChunkFuture from futureMap as soon as it completes.
was (Author: shashikant):
Thanks [~xyao], for the review comments. Patch v3 fixes the compilation failure. The idea is to add the writeChunkFuture in the futureMap and once it finishes should remove it from the Map. Patch v3 has the correct ordering.
> Serialize ApplyTransaction calls per Container in ContainerStateMachine
> -----------------------------------------------------------------------
>
> Key: HDDS-550
> URL: https://issues.apache.org/jira/browse/HDDS-550
> Project: Hadoop Distributed Data Store
> Issue Type: Bug
> Reporter: Shashikant Banerjee
> Assignee: Shashikant Banerjee
> Priority: Major
> Attachments: HDDS-550.001.patch, HDDS-550.002.patch, HDDS-550.003.patch
>
>
> As part of handling Node failures in Ozone, the block commit need to happen in order inside ContainerStateMachine per container. With RATIS-341, it is guaranteed that the applyTransaction calls for committing the write chunks will be initiated only when the WriteStateMachine data for write Chunk operations finish.
> This Jira is aimed at making all the applyTransaction operations inside ContainerStateMachine serial per container with a single thread Executor per container handling all applyTransactions calls.
--
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