hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Chen Liang (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HDFS-13767) Add msync server implementation.
Date Thu, 26 Jul 2018 21:50:00 GMT

    [ https://issues.apache.org/jira/browse/HDFS-13767?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16558934#comment-16558934

Chen Liang commented on HDFS-13767:

Upload WIP.002 patch. The main difference from WIP.001 patch is that, the logic to make sure
the calls from the same client stays the  same processing order is removed. Specifically,
if a call has state id large than server state id, the Handler will simply insert the call
back to the callQueue and continue. As an example, say callQueue has two calls from same client [1,2].
1 gets checked, and the server state id hasn't caught up. Then 1 gets added back to queue,
making it [2, 1]. Then server caught up to state id, say, 3. Then 2 gets checked, and processed,
then 1. So the processing order becomes 2,1.

But this is fine because even in current Server logic, there is no guarantee on the order:
It is already possible that two handler threads pick up 1 and 2 respectively and 2 finishes
first. In fact, due to the synchronized natural of the API, only when the same client instance
is used by multiple threads, there will be multiple calls from the same client in the callQueue.
But in this case, there should be no expectation on ordering. Furthermore, this logic is
for Observer exclusively, which only handles reads. (Please correct me if I'm wrong on this).

> Add msync server implementation.
> --------------------------------
>                 Key: HDFS-13767
>                 URL: https://issues.apache.org/jira/browse/HDFS-13767
>             Project: Hadoop HDFS
>          Issue Type: Sub-task
>          Components: namenode
>            Reporter: Chen Liang
>            Assignee: Chen Liang
>            Priority: Major
>         Attachments: HDFS-13767.WIP.001.patch, HDFS-13767.WIP.002.patch
> This is a followup on HDFS-13688, where msync API is introduced to {{ClientProtocol}}
but the server side implementation is missing. This is Jira is to implement the server side

This message was sent by Atlassian JIRA

To unsubscribe, e-mail: hdfs-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-help@hadoop.apache.org

View raw message