saxenapranav opened a new pull request, #6763:
URL: https://github.com/apache/hadoop/pull/6763

   JIRA: https://issues.apache.org/jira/browse/HADOOP-19102
   PR on trunk: https://github.com/apache/hadoop/pull/6617
   Merged commit on trunk: 
https://github.com/apache/hadoop/commit/6404692c0973a7b018ca77f4aaad4248b62782e2
   
   The method `optimisedRead` creates a buffer array of size `readBufferSize`. 
If footerReadBufferSize is greater than readBufferSize, abfs will attempt to 
read more data than the buffer array can hold, which causes an exception.
   
   Change: To avoid this, we will keep footerBufferSize = 
min(readBufferSizeConfig, footerBufferSizeConfig)
   Test change: `ITestAbfsInputStreamReadFooter` tests different scenarios with 
different combinations of fileSize and footerBufferReadSize. Have added a 
dimension of readBufferSize in the testcases. Now its a combination of 
fileSize, readBufferSize, footerBufferReadSize.
   
   
   Also, as part of this PR, have improved tests within 
`ITestAbfsInputStreamReadFooter`. There are tests which have multiple 
combination, and there was file getting created for all the combination. There 
has to be a combination on different fileSize. 
   The change: We will spin up one thread each for each fileSize. And in each 
thread, all the combination for that particular fileSize will run. This will 
help in creating file once for a fileSize and multiple fileSize related 
assertion can happen in parallel and use hardware capability.
   Improvement: on a 6 processor VM [outside Azure network], on trunk, it tool 
8min47sec to run all tests of ITestAbfsInputStreamReadFooter and in the PR 
branch, it took 7 min. (Its IDE run wherein each test method run one after 
another unlike sunfire-maven command(used in runTest script) which can run 
tests in parallel).
   


-- 
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: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


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

Reply via email to