hadoop-common-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Steve Loughran (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HADOOP-12444) Consider implementing lazy seek in S3AInputStream
Date Fri, 05 Feb 2016 21:55:39 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-12444?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15135090#comment-15135090
] 

Steve Loughran commented on HADOOP-12444:
-----------------------------------------

Thanks for the patch on EOF

FWIW, you'd have caught this problem on running the S3a contract tests —and been able to
turn it off by editing the contract-test options.xml. Some other filesystems (real posix ones)
don't faill on a seek past the end, but only on the read (they have to allow for a write).
Anyway, what you have done at least makes it consistent with HDFS.

However, I do have a strict policy regarding testing here, namely you have to have run it
yourself against at least one AWS instance. And I have to insistt on that for everyone to
avoid being seen as relaxing the criteria for some patches https://wiki.apache.org/hadoop/HowToContribute#Submitting_patches_against_object_stores_such_as_Amazon_S3.2C_OpenStack_Swift_and_Microsoft_Azure

Have you done it yet, and if so which one?

thanks

> Consider implementing lazy seek in S3AInputStream
> -------------------------------------------------
>
>                 Key: HADOOP-12444
>                 URL: https://issues.apache.org/jira/browse/HADOOP-12444
>             Project: Hadoop Common
>          Issue Type: Sub-task
>          Components: fs/s3
>    Affects Versions: 2.7.1
>            Reporter: Rajesh Balamohan
>            Assignee: Rajesh Balamohan
>         Attachments: HADOOP-12444.1.patch, HADOOP-12444.2.patch, HADOOP-12444.WIP.patch
>
>
> - Currently, "read(long position, byte[] buffer, int offset, int length)" is not implemented
in S3AInputStream (unlike DFSInputStream). So, "readFully(long position, byte[] buffer, int
offset, int length)" in S3AInputStream goes through the default implementation of seek(),
read(), seek() in FSInputStream. 
> - However, seek() in S3AInputStream involves re-opening of connection to S3 everytime
(https://github.com/apache/hadoop/blob/trunk/hadoop-tools/hadoop-aws/src/main/java/org/apache/hadoop/fs/s3a/S3AInputStream.java#L115).
 
> - It would be good to consider having a lazy seek implementation to reduce connection
overheads to S3. (e.g Presto implements lazy seek. https://github.com/facebook/presto/blob/master/presto-hive/src/main/java/com/facebook/presto/hive/PrestoS3FileSystem.java#L623)



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message