hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Yiqun Lin (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (HDFS-11580) Ozone: Support asynchronus client API for SCM and containers
Date Wed, 03 May 2017 05:26:04 GMT

     [ https://issues.apache.org/jira/browse/HDFS-11580?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Yiqun Lin updated HDFS-11580:
-----------------------------
    Attachment: HDFS-11580-HDFS-7240.006.patch

Hi [~msingh], thanks for your review! The comments you suggested make sense to me. The following
are my comments.

{quote}
1) Thanks for dividing the apis into sync and async. I feel that for sync apis, lets use the
previous function without any change. This is to ensure that current behavior doesn't break.
{quote}
Actually I didn't change any places in the original sync apis (including the params, method
name and logic). Just some code format. You can have a check, :).
{quote}
Also can we add an extra capability to {{ XceiverClientHandler }} which initializes it as
either sync or async. Only a sync client will be able to send sync command and vice versa.
{quote}
I have a little different option on this. Here the async way in {{XceiverClientHandler}} is
transparent for client/users. So Is this necessary to do?
{quote}
3) in {{ testCorrectnessWithMultipleAsyncCalls }}, I feel that the return future should be
accepted in an array and then they can waited upon in another loop.This will ensure that commands
are enqueued in parallel
{quote}
Good catch! Addressed in the latest patch.
{quote}
4) Another comment which has earlier been noted, this code will not work with cblock correctly.
as there are cases where TraceID is not set correctly.
{quote}
Sorry for missing your cblock codes. One quick fix way I am imaging is that we can do an additional
check for the traceID, if the id is empty ("" or null), we should give it a unique id (here
I use UUID.randomUUID()). But for long-term consideration, this way should be improved. It's
not a good way to generate id everywhere. It would be better to use a special id generator
to help us generate unique id. I will file a new ticket to for this.

Attach the updated patch.

> Ozone: Support asynchronus client API for SCM and containers
> ------------------------------------------------------------
>
>                 Key: HDFS-11580
>                 URL: https://issues.apache.org/jira/browse/HDFS-11580
>             Project: Hadoop HDFS
>          Issue Type: Sub-task
>          Components: ozone
>    Affects Versions: HDFS-7240
>            Reporter: Anu Engineer
>            Assignee: Yiqun Lin
>         Attachments: HDFS-11580-HDFS-7240.001.patch, HDFS-11580-HDFS-7240.002.patch,
HDFS-11580-HDFS-7240.003.patch, HDFS-11580-HDFS-7240.004.patch, HDFS-11580-HDFS-7240.005.patch,
HDFS-11580-HDFS-7240.006.patch
>
>
> This is an umbrella JIRA that needs to support a set of APIs in Asynchronous form.
> For containers -- or the datanode API currently supports a call {{sendCommand}}. we need
to build proper programming interface and support an async interface.
> There is also a set of SCM API that clients can call, it would be nice to support Async
interface for those too.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

---------------------------------------------------------------------
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