[
https://issues.apache.org/jira/browse/HADOOP-12304?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Chris Nauroth updated HADOOP-12304:
-----------------------------------
Attachment: HADOOP-12304.001.patch
The attached patch changes {{DelegateToFileSystem}} to translate the 0 returned
from {{FileSystem#getDefaultPort}} to the -1 expected by
{{AbstractFileSystem#getUri}}. I verified that the tests written for
HADOOP-11618 still pass. I also added a new test to confirm that using wasb as
the default file system uses a correct URL (without port :0) when accessed
through {{FileContext}}. I also have done manual testing to confirm that I can
run MapReduce jobs with wasb as the default file system after this change.
> Applications using FileContext fail with the default file system configured
> to be wasb/s3/etc.
> ----------------------------------------------------------------------------------------------
>
> Key: HADOOP-12304
> URL: https://issues.apache.org/jira/browse/HADOOP-12304
> Project: Hadoop Common
> Issue Type: Bug
> Components: fs
> Affects Versions: 2.7.0, 2.7.1
> Reporter: Chris Nauroth
> Assignee: Chris Nauroth
> Priority: Blocker
> Labels: 2.7.2-candidate
> Attachments: HADOOP-12304.001.patch
>
>
> HADOOP-11618 fixed a bug with {{DelegateToFileSystem}} using the wrong
> default port. As a side effect of this patch, file path URLs that previously
> had no port now insert :0 for the port, as per the default implementation of
> {{FileSystem#getDefaultPort}}. At runtime, this can cause an application to
> erroneously try contacting port 0 for a remote blob store service. The
> connection fails. Ultimately, this renders wasb, s3, and probably custom
> file system implementations outside the Hadoop source tree completely
> unusable as the default file system.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
