[ 
https://issues.apache.org/jira/browse/HADOOP-14738?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Steve Loughran updated HADOOP-14738:
------------------------------------
    Description: 
We are all happy with S3A; it's been stable since Hadoop 2.7 and high-perf 
since Hadoop 2.8

It's now time to kill S3N off, remove the source, the tests, the transitive 
dependencies. This patch does that.

It also removes the obsolete, original s3a output stream; the fast/block upload 
stream has been stable and is much more manageable and maintained (put 
differently: we don't ever look at the original S3A output stream, and tell 
people not to use it for performance reasons).

As well as cutting the features, this patch updates the aws docs with
* split out s3n migration page
* split out troubleshooting page
* rework of the "uploading data with s3a" section of index.md, as there's no 
need to discuss the slow upload except in the past tense...all that is needed 
is to list the buffering and thread tuning options of the block uploader.


  was:
We are all happy with S3A; it's been stable since Hadoop 2.7 and high-perf 
since Hadoop 2.8

It's now time to kill S3N off, remove the source, the tests, the transitive 
dependencies.

I propose that in Hadoop 3.0 beta we tell people off from using it, and link to 
a doc page (wiki?) about how to migrate (Change URLs, update config ops).




> Remove S3N and obsolete bits of S3A; rework docs
> ------------------------------------------------
>
>                 Key: HADOOP-14738
>                 URL: https://issues.apache.org/jira/browse/HADOOP-14738
>             Project: Hadoop Common
>          Issue Type: Sub-task
>          Components: fs/s3
>    Affects Versions: 2.9.0, 3.0.0-beta1
>            Reporter: Steve Loughran
>            Assignee: Steve Loughran
>            Priority: Blocker
>         Attachments: HADOOP-14738-002.patch, HADOOP-14738-003.patch, 
> HADOOP-14738-004.patch, HADOOP-14739-001.patch
>
>
> We are all happy with S3A; it's been stable since Hadoop 2.7 and high-perf 
> since Hadoop 2.8
> It's now time to kill S3N off, remove the source, the tests, the transitive 
> dependencies. This patch does that.
> It also removes the obsolete, original s3a output stream; the fast/block 
> upload stream has been stable and is much more manageable and maintained (put 
> differently: we don't ever look at the original S3A output stream, and tell 
> people not to use it for performance reasons).
> As well as cutting the features, this patch updates the aws docs with
> * split out s3n migration page
> * split out troubleshooting page
> * rework of the "uploading data with s3a" section of index.md, as there's no 
> need to discuss the slow upload except in the past tense...all that is needed 
> is to list the buffering and thread tuning options of the block uploader.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to