flink-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "mingleizhang (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (FLINK-6381) Unnecessary synchronizing object in BucketingSink
Date Thu, 04 May 2017 07:43:04 GMT

     [ https://issues.apache.org/jira/browse/FLINK-6381?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

mingleizhang updated FLINK-6381:
--------------------------------
    Description: 
It seems that currently there are two places should not employ the {{synchronized}} to describe
{{pendingFilesPerCheckpoint}}, as it is only restored state object for checkpoint and no sharing
of the data-structure between different threads, as follows. {code} private void handleRestoredRollingSinkState(RollingSink.BucketState
restoredState) {
     ...
    synchronized (restoredState.pendingFilesPerCheckpoint) {
			restoredState.pendingFilesPerCheckpoint.clear();
		}
     ...
}{code} and {code}private void handleRestoredBucketState(State<T> restoredState) { 
     
    ...
    synchronized (bucketState.pendingFilesPerCheckpoint) {
	     	        bucketState.pendingFilesPerCheckpoint.clear();
		}
 } {code}
Hi, [~kkl0u]. Is there any other stuff shoud add here ? Would you mind have a more thorough
look in this class ? Thanks go out to you. I am very appreciate it.

  was:
It seems that currently there are two places should not employ the {{synchronized}} to describe
{{pendingFilesPerCheckpoint}}, as it is only restored state object for checkpoint and no sharing
of the data-structure between different threads. As follows. {code} private void handleRestoredRollingSinkState(RollingSink.BucketState
restoredState) {
     ...
    synchronized (restoredState.pendingFilesPerCheckpoint) {
			restoredState.pendingFilesPerCheckpoint.clear();
		}
     ...
}{code} and {code}private void handleRestoredBucketState(State<T> restoredState) { 
     
    ...
    synchronized (bucketState.pendingFilesPerCheckpoint) {
	     	        bucketState.pendingFilesPerCheckpoint.clear();
		}
 } {code}
Hi, [~kkl0u]. Is there any other stuff shoud add here ? Would you mind have a more thorough
look in this class ? Thanks go out to you. I am very appreciate it.


> Unnecessary synchronizing object in BucketingSink
> -------------------------------------------------
>
>                 Key: FLINK-6381
>                 URL: https://issues.apache.org/jira/browse/FLINK-6381
>             Project: Flink
>          Issue Type: Improvement
>          Components: filesystem-connector
>            Reporter: mingleizhang
>            Assignee: mingleizhang
>
> It seems that currently there are two places should not employ the {{synchronized}} to
describe {{pendingFilesPerCheckpoint}}, as it is only restored state object for checkpoint
and no sharing of the data-structure between different threads, as follows. {code} private
void handleRestoredRollingSinkState(RollingSink.BucketState restoredState) {
>      ...
>     synchronized (restoredState.pendingFilesPerCheckpoint) {
> 			restoredState.pendingFilesPerCheckpoint.clear();
> 		}
>      ...
> }{code} and {code}private void handleRestoredBucketState(State<T> restoredState)
{       
>     ...
>     synchronized (bucketState.pendingFilesPerCheckpoint) {
> 	     	        bucketState.pendingFilesPerCheckpoint.clear();
> 		}
>  } {code}
> Hi, [~kkl0u]. Is there any other stuff shoud add here ? Would you mind have a more thorough
look in this class ? Thanks go out to you. I am very appreciate it.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

Mime
View raw message