[
https://issues.apache.org/jira/browse/HADOOP-11584?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14329721#comment-14329721
]
Lei (Eddy) Xu commented on HADOOP-11584:
----------------------------------------
Tests passed on AWS US standard region. Non-binding +1.
Just one small question, why do {{TestS3ABlocksize}} and
{{TestS3AFileSystemContract}} require different {{s3a fs name}} ?
{{TestS3ABlocksize}} asks for {{fs.contract.test.fs.s3a}} while
{{TestS3AFileSystemContract}} asks for {{test.fs.s3a.name}}?
> s3a file block size set to 0 in getFileStatus
> ---------------------------------------------
>
> Key: HADOOP-11584
> URL: https://issues.apache.org/jira/browse/HADOOP-11584
> Project: Hadoop Common
> Issue Type: Sub-task
> Components: fs/s3
> Affects Versions: 2.6.0
> Reporter: Dan Hecht
> Assignee: Brahma Reddy Battula
> Priority: Blocker
> Attachments: HADOOP-10584-003.patch, HADOOP-111584.patch,
> HADOOP-11584-002.patch
>
>
> The consequence is that mapreduce probably is not splitting s3a files in the
> expected way. This is similar to HADOOP-5861 (which was for s3n, though s3n
> was passing 5G rather than 0 for block size).
> FileInputFormat.getSplits() relies on the FileStatus block size being set:
> {code}
> if (isSplitable(job, path)) {
> long blockSize = file.getBlockSize();
> long splitSize = computeSplitSize(blockSize, minSize, maxSize);
> {code}
> However, S3AFileSystem does not set the FileStatus block size field. From
> S3AFileStatus.java:
> {code}
> // Files
> public S3AFileStatus(long length, long modification_time, Path path) {
> super(length, false, 1, 0, modification_time, path);
> isEmptyDirectory = false;
> }
> {code}
> I think it should use S3AFileSystem.getDefaultBlockSize() for each file's
> block size (where it's currently passing 0).
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
