hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ajay Kumar (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HDDS-187) Command status publisher for datanode
Date Thu, 12 Jul 2018 07:27:00 GMT

    [ https://issues.apache.org/jira/browse/HDDS-187?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16541256#comment-16541256
] 

Ajay Kumar commented on HDDS-187:
---------------------------------

[~anu] thanks for review. Addressed most of your comments in patch v10.
{quote}CommandStatus.java:32*: rewrite as private Status status;
Is it possible to add cmdID to base SCMCommand Class instead of adding it to all derived classes
?
{quote}
done.
{quote}Same comment on Handle Call, may be make an abstract class that deal with the status
and ID ?{quote}
done. Default method in CommandHandler interface takes few parameters but does the job.
{quote}Do we need HDDsIDFactory at all from quick reading of code it looks like all we need
is Time.MonotonicNow(){quote}
Initial patches were with {{Time.MonotonicNow()}} but [~xyao][# review comment|https://issues.apache.org/jira/browse/HDDS-187?focusedCommentId=16533982&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-16533982]
suggested having a factory method which can be reused in other places as well. I thought its
good idea and introduced this API. Although current approach is mixed version of that as we
are not persisting last unique id used. (to restart from there, but IMO current use case is
not effected by it)
{quote}Why Skip Reregister command in HeartbeatTask.java ? {quote}
If node re-registered fails than tracking its command status doesn't help as it will not be
reported back. If it is re-registered successfully than also command status of REREGISTER
is not of any value as node already reregistered succesfully. Also it has no handler to set
its status as executed.
{quote}Should we do the addCommandStatus() inside addCommand() ?{quote}
done

> Command status publisher for datanode
> -------------------------------------
>
>                 Key: HDDS-187
>                 URL: https://issues.apache.org/jira/browse/HDDS-187
>             Project: Hadoop Distributed Data Store
>          Issue Type: Bug
>          Components: SCM
>    Affects Versions: 0.2.1
>            Reporter: Ajay Kumar
>            Assignee: Ajay Kumar
>            Priority: Major
>             Fix For: 0.2.1
>
>         Attachments: HDDS-187.00.patch, HDDS-187.01.patch, HDDS-187.02.patch, HDDS-187.03.patch,
HDDS-187.04.patch, HDDS-187.05.patch, HDDS-187.06.patch, HDDS-187.07.patch, HDDS-187.08.patch,
HDDS-187.09.patch, HDDS-187.10.patch
>
>
> Currently SCM sends set of commands for DataNode. DataNode executes them via CommandHandler.
This jira intends to create a Command status publisher which will return status of these commands
back to the SCM.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

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


Mime
View raw message