[
https://issues.apache.org/jira/browse/HDFS-16620?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Yongpan Liu updated HDFS-16620:
---
Description: ClientProtocol#msync is invoked frequently in clusters where
the Observer is enabled. When NameNode is overwhelmed with load, the average
response time of msync will increase, which can affect Observer Read
efficiency. We can split out a separate RPC server just like HDFS-9311. (was:
ClientProtocol#msync is invoked frequently in clusters where the Observer is
enabled. When NameNode is heavily loaded, the average response time of msync
increases, which can affect Observer Read efficiency. We can split out a
separate RPC server just like HDFS-9311.)
> Improve msync performance by a separate RPC server
> --
>
> Key: HDFS-16620
> URL: https://issues.apache.org/jira/browse/HDFS-16620
> Project: Hadoop HDFS
> Issue Type: Improvement
>Reporter: Yongpan Liu
>Priority: Minor
>
> ClientProtocol#msync is invoked frequently in clusters where the Observer is
> enabled. When NameNode is overwhelmed with load, the average response time of
> msync will increase, which can affect Observer Read efficiency. We can split
> out a separate RPC server just like HDFS-9311.
--
This message was sent by Atlassian Jira
(v8.20.7#820007)
-
To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org