[
https://issues.apache.org/jira/browse/HADOOP-14140?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15896251#comment-15896251
]
Steve Loughran commented on HADOOP-14140:
-----------------------------------------
don't worry about spark for now, it will only complicate your life. Replicate
this in a snippet of a junit test or with the 'hadoop fs' command.
Once you are confident it works, then you could look at spark
# you can't mix them unless you replace every hadoop-* JAR
# if you look at SPARK-7481 you can get the PR I use to build this stuff
However, before you bother, consider this. There is no point running any
analytics query against an empty bucket. There's no data, there's no answer.
It's a corner case, which is why it didn't surface for a while. Prioritise your
problems.
> S3A reporting "Not found on empty bucket
> ----------------------------------------
>
> Key: HADOOP-14140
> URL: https://issues.apache.org/jira/browse/HADOOP-14140
> Project: Hadoop Common
> Issue Type: Bug
> Components: fs/s3
> Affects Versions: 2.7.3
> Environment: Non-AWS s3 implementation
> Reporter: Vishnu Vardhan
> Priority: Minor
>
> Hi:
> UPDATE: This stack trace is being caused because there are 0 objects in the
> target directory. S3A is unable to handle 0 objects
> ---
> Connecting S3A to a 3rd party object store does not work. This is a publicly
> hosted grid and i can provide credentials if required. Please see the debug
> log below
> There are two problems -
> 1. Path Style setting is ignored, and S3A always uses host style addressing
> 2. Even when host style is specified, it is unable to proceed, see debug log
> (in comment below)
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
