[
https://issues.apache.org/jira/browse/HADOOP-12537?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14985024#comment-14985024
]
Steve Loughran commented on HADOOP-12537:
-----------------------------------------
s3: has been essentially deprecated for a long time, nobody uses it for new data
s3n is considered stable, and after the disaster that was HADOOP-9623, in which
a patch which swallowed exceptions snuck in, there's effectively a veto on any
patch there that isn't considered critical in terms of security or performance.
No new features.
The failures that patch triggered (HADOOP-10589) showed that there wasn 't
adequate testing of the s3 clients; that's better now, but it means there'll be
an expectation of more tests for any feature; tests that are (a) robust even
over long-haul connections (b) skip nicely against non-AWS implementations of
the S3 APIs and (c) tested by you against multiple s3 endpoints, including the
more recent ones with tighter auth (e.g. AWS frankfurt).
> s3a: Add flag for session ID to allow Amazon STS temporary credentials
> ----------------------------------------------------------------------
>
> Key: HADOOP-12537
> URL: https://issues.apache.org/jira/browse/HADOOP-12537
> Project: Hadoop Common
> Issue Type: New Feature
> Components: fs/s3
> Affects Versions: 2.7.1
> Reporter: Sean Mackrory
> Priority: Minor
>
> Amazon STS allows you to issue temporary access key id / secret key pairs for
> your a user / role. However, using these credentials also requires specifying
> a session ID. There is currently no such configuration property or the
> required code to pass it through to the API (at least not that I can find) in
> any of the S3 connectors.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
