[ 
https://issues.apache.org/jira/browse/HADOOP-18236?focusedWorklogId=774755&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-774755
 ]

ASF GitHub Bot logged work on HADOOP-18236:
-------------------------------------------

                Author: ASF GitHub Bot
            Created on: 25/May/22 18:24
            Start Date: 25/May/22 18:24
    Worklog Time Spent: 10m 
      Work Description: ayushtkn commented on PR #4320:
URL: https://github.com/apache/hadoop/pull/4320#issuecomment-1137686111

   Logically from the present code flow this lock is redundant and there is 
enough discussion on the original jira, this was kept because the method 
calling was protected and a comment there confirms that it is indeed deliberate
   
https://issues.apache.org/jira/browse/HADOOP-16028?focusedCommentId=16731433&page=com.atlassian.jira.plugin.system.issuetabpanels%3Acomment-tabpanel#comment-16731433
   
   I don't think we can change this access modifier to private for compat 
reasons and the reason for which it was done then still holds true




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

    Worklog Id:     (was: 774755)
    Time Spent: 1h 20m  (was: 1h 10m)

> Remove duplicate locks in NetworkTopology
> -----------------------------------------
>
>                 Key: HADOOP-18236
>                 URL: https://issues.apache.org/jira/browse/HADOOP-18236
>             Project: Hadoop Common
>          Issue Type: Improvement
>            Reporter: ZanderXu
>            Assignee: ZanderXu
>            Priority: Minor
>              Labels: pull-request-available
>          Time Spent: 1h 20m
>  Remaining Estimate: 0h
>
> During reading the hadoop NetworkTopology.java, I suspect there is a 
> duplicate lock.
> chooseRandom(line 532), and code is:
> {code:java}
> final int availableNodes;
> if (excludedScope == null) {
>   availableNodes = countNumOfAvailableNodes(scope, excludedNodes);
> } else {
>   netlock.readLock().lock();
>   try {
>     availableNodes = countNumOfAvailableNodes(scope, excludedNodes) -
>         countNumOfAvailableNodes(excludedScope, excludedNodes);
>   } finally {
>     netlock.readLock().unlock();
>   }
> } {code}
> All the places where called `chooseRandom` have the global read lock, so the 
> internal read lock is duplicated.



--
This message was sent by Atlassian Jira
(v8.20.7#820007)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to