[ 
https://issues.apache.org/jira/browse/HADOOP-18186?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17584501#comment-17584501
 ] 

ASF GitHub Bot commented on HADOOP-18186:
-----------------------------------------

virajjasani commented on PR #4796:
URL: https://github.com/apache/hadoop/pull/4796#issuecomment-1226578101

   @steveloughran i also tested the entire test suite by enabling prefatch. 
Tests in `ITestS3AContractVectoredRead` failed. One example:
   ```
   [ERROR] testMinSeekAndMaxSizeDefaultValues[Buffer type : 
direct](org.apache.hadoop.fs.contract.s3a.ITestS3AContractVectoredRead)  Time 
elapsed: 13.025 s  <<< FAILURE!
   org.junit.ComparisonFailure: [Mismatch in default s3a min seek for vectored 
reads] expected:<4[8]96> but was:<4[0]96>
        at sun.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method)
        at 
sun.reflect.NativeConstructorAccessorImpl.newInstance(NativeConstructorAccessorImpl.java:62)
        at 
sun.reflect.DelegatingConstructorAccessorImpl.newInstance(DelegatingConstructorAccessorImpl.java:45)
        at org.apache.hadoop.test.MoreAsserts.assertEqual(MoreAsserts.java:99)
        at 
org.apache.hadoop.fs.contract.s3a.ITestS3AContractVectoredRead.testMinSeekAndMaxSizeDefaultValues(ITestS3AContractVectoredRead.java:111)
   
   ```
   
   However, the failures don't seem relevant to the change on this PR. As per 
the test logic, it seems`ITestS3AContractVectoredRead` failures are expected 
when prefetch is enabled.




> s3a prefetching to use SemaphoredDelegatingExecutor for submitting work
> -----------------------------------------------------------------------
>
>                 Key: HADOOP-18186
>                 URL: https://issues.apache.org/jira/browse/HADOOP-18186
>             Project: Hadoop Common
>          Issue Type: Sub-task
>          Components: fs/s3
>    Affects Versions: 3.4.0
>            Reporter: Steve Loughran
>            Assignee: Viraj Jasani
>            Priority: Major
>              Labels: pull-request-available
>
> Use SemaphoredDelegatingExecutor for each to stream to submit work, if 
> possible, for better fairness in processes with many streams.
> this also takes a DurationTrackerFactory to count how long was spent in the 
> queue, something we would want to know



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

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

Reply via email to