[
https://issues.apache.org/jira/browse/HDFS-16088?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17372119#comment-17372119
]
Xiaoqiao He commented on HDFS-16088:
------------------------------------
Thanks [~tomscut] involve me here. It makes sense to me at first glance.
However #getLiveDatanodeStorageReport is not quite the same with #getBlocks
method IMO, especially it is also used by mover and diskbalancer, I am not sure
if it is graceful to request redirect to Standby on HA mode. + It will be more
helpful to extent the unit test. Thanks.
> Standby NameNode process getLiveDatanodeStorageReport request to reduce
> Active load
> -----------------------------------------------------------------------------------
>
> Key: HDFS-16088
> URL: https://issues.apache.org/jira/browse/HDFS-16088
> Project: Hadoop HDFS
> Issue Type: Improvement
> Reporter: tomscut
> Assignee: tomscut
> Priority: Minor
> Labels: pull-request-available
> Time Spent: 0.5h
> Remaining Estimate: 0h
>
> As with [HDFS-13183|https://issues.apache.org/jira/browse/HDFS-13183],
> NameNodeConnector#getLiveDatanodeStorageReport() can also request to SNN to
> reduce the ANN load.
> There are two points that need to be mentioned:
> 1. NameNodeConnector#getLiveDatanodeStorageReport() is
> OperationCategory.UNCHECKED in FSNamesystem, so we can access SNN directly.
> 2. We can share the same UT(testBalancerRequestSBNWithHA) with
> NameNodeConnector#getBlocks().
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
