[jira] [Work logged] (GOBBLIN-1619) WriterUtils.mkdirsWithRecursivePermission contains race condition and puts unnecessary load on filesystem

["ASF GitHub Bot (Jira)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/GOBBLIN-1619?focusedWorklogId=739784&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-739784 ]

ASF GitHub Bot logged work on GOBBLIN-1619:
-------------------------------------------

                Author: ASF GitHub Bot
            Created on: 11/Mar/22 00:40
            Start Date: 11/Mar/22 00:40
    Worklog Time Spent: 10m 
      Work Description: homatthew opened a new pull request #3477:
URL: https://github.com/apache/gobblin/pull/3477


   Dear Gobblin maintainers,
   
   Please accept this PR. I understand that it will not be reviewed until I have checked off all the steps below!
   
   
   ### JIRA
   - [Z] My PR addresses the following [Gobblin JIRA](https://issues.apache.org/jira/browse/GOBBLIN/) issues and references them in the PR title. For example, "[GOBBLIN-1619] My Gobblin PR"
       - https://issues.apache.org/jira/browse/GOBBLIN-1619
   
   
   ### Description
   - [X] Here are some details about my PR, including screenshots (if applicable):
   #### The current implementation recursively calls fs.mkdirs has the following issues:
   - Race condition for creating parent directories, causing FileNotFound exception even when the file exists on file system
   - HDFS fs.mkdirs atomically creates missing parent directories. Thus, the recursive approach is making unnecessary calls.
   
   In HDFS, which the current [FileSystem](https://hadoop.apache.org/docs/stable/api/org/apache/hadoop/fs/FileSystem.html) interface is built upon, guarantees the parents will be created. So, all FileSystem class implementations should also follow this behavior. 
   
    
   
   See the [FileSystem](https://hadoop.apache.org/docs/stable/api/org/apache/hadoop/fs/FileSystem.html) abstract class documentation:
   
   > The behaviour of the filesystem is [specified in the Hadoop documentation. ](https://hadoop.apache.org/docs/stable/hadoop-project-dist/hadoop-common/filesystem/filesystem.html)However, the normative specification of the behavior of this class is actually HDFS: if HDFS does not behave the way these Javadocs or the specification in the Hadoop documentations define, assume that the documentation is incorrect
   
   ### Tests
   - [X] My PR adds the following unit tests __OR__ does not need testing for this extremely good reason:
   Unit tests added and change tested internally at LinkedIn on parallel fastingest test pipeline. The parallel test pipeline compared the permissions created with previous implementation and with new implementation. There were not any differences in the way permissions are set, so the change is completely transparent and non-breaking. 
   
   ### Commits
   - [X] My commits all reference JIRA issues in their subject lines, and I have squashed multiple commits if they address the same issue. In addition, my commits follow the guidelines from "[How to write a good git commit message](http://chris.beams.io/posts/git-commit/)":
       1. Subject is separated from body by a blank line
       2. Subject is limited to 50 characters
       3. Subject does not end with a period
       4. Subject uses the imperative mood ("add", not "adding")
       5. Body wraps at 72 characters
       6. Body explains "what" and "why", not "how"
   
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: dev-unsubscribe@gobblin.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


Issue Time Tracking
-------------------

            Worklog Id:     (was: 739784)
    Remaining Estimate: 0h
            Time Spent: 10m

> WriterUtils.mkdirsWithRecursivePermission contains race condition and puts unnecessary load on filesystem
> ---------------------------------------------------------------------------------------------------------
>
>                 Key: GOBBLIN-1619
>                 URL: https://issues.apache.org/jira/browse/GOBBLIN-1619
>             Project: Apache Gobblin
>          Issue Type: Bug
>            Reporter: Matthew Ho
>            Priority: Minor
>          Time Spent: 10m
>  Remaining Estimate: 0h
>
> The current implementation recursively calls fs.mkdirs has the following issues:
>  * *Race condition for creating parent directories, causing FileNotFound exception even when the file exists on file system*
>  * {*}HDFS fs.mkdirs atomically creates missing parent directories. Thus, the recursive approach is making unnecessary calls.{*}{*}{*}
> HDFS, which the current FileSystem interface is built upon, guarantees the parents will be created. So all FileSystem class implementations should also follow this behavior. 
>  
> *Note the [FileSystem|https://hadoop.apache.org/docs/stable/api/org/apache/hadoop/fs/FileSystem.html] abstract class documentation says the following:*
> The behaviour of the filesystem is [specified in the Hadoop documentation. |https://hadoop.apache.org/docs/stable/hadoop-project-dist/hadoop-common/filesystem/filesystem.html]However, the normative specification of the behavior of this class is actually HDFS: {color:#de350b}if HDFS does not behave the way these Javadocs or the specification in the Hadoop documentations define, assume that the documentation is incorrect{color}



--
This message was sent by Atlassian Jira
(v8.20.1#820001)