[
https://issues.apache.org/jira/browse/HDFS-14710?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16905170#comment-16905170
]
Yuxuan Wang commented on HDFS-14710:
------------------------------------
Before the patch, if I change/add a mount table entry which involving some
opening files, clients may get some unexpected exceptions. e.g. client calls
*complete* but the destination has changed, we should resolve to original path
not the new.
And here, *abandonBlock*, *updateBlockForPipeline*, *updatePipeline* just use
parameter *extendedBlock* to determine the dest.
I mean, I think it's not just Improve some RPC performances.
> RBF: Improve some RPC performances
> ----------------------------------
>
> Key: HDFS-14710
> URL: https://issues.apache.org/jira/browse/HDFS-14710
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: rbf
> Reporter: xuzq
> Assignee: xuzq
> Priority: Minor
> Attachments: HDFS-14710-trunk-001.patch, HDFS-14710-trunk-002.patch,
> HDFS-14710-trunk-003.patch
>
>
> We can improve some RPC performance if the extendedBlock is not null.
> Such as addBlock, getAdditionalDatanode and complete.
> Since HDFS encourages user to write large files, so the extendedBlock is not
> null in most case.
> In the scenario of Multiple Destination and large file, the effect is more
> obvious.
--
This message was sent by Atlassian JIRA
(v7.6.14#76016)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
