[
https://issues.apache.org/jira/browse/HDDS-3778?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17133110#comment-17133110
]
Shashikant Banerjee commented on HDDS-3778:
-------------------------------------------
The solution is to keep block allocation synchronised on the pipeline but it
leads to perf degradation by 50% as indicated by Genesis benchmark for allocate
block.
Thanks [~nanda] for the benchmark results.
> Block distribution in a pipeline among open containers is not uniform
> ---------------------------------------------------------------------
>
> Key: HDDS-3778
> URL: https://issues.apache.org/jira/browse/HDDS-3778
> Project: Hadoop Distributed Data Store
> Issue Type: Bug
> Components: SCM
> Reporter: Shashikant Banerjee
> Assignee: Shashikant Banerjee
> Priority: Blocker
> Fix For: 0.7.0
>
> Attachments: With-fully-synchronized-getMatchingContainer.png,
> Without-fully-synchronized-getMatchingContainer.png
>
>
> Currently, with concurrent allocate block calls, the block allocation among
> the open containers of a pipeline is not uniform as with concurrent logic,
> block allocation logic with last used notion does not hold up. The idea here
> is to address this.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
