Akshat-Jain commented on PR #16481:
URL: https://github.com/apache/druid/pull/16481#issuecomment-2134806610

   @rohangarg 
   
   > Ok, although the bandwidth comes out to be around < 1 MB/sec with this 
calculation. So, I'm not sure about realistic nature of the test.
   
   The data I mentioned was average of 3+ readings. I can do it again (once all 
review comments are addressed and everyone is okay with the implementation).
   
   > Now, if you've written a chunk, you'll put it in an executor which might 
be consumed by the parts of other writers and you'll wait in the queue for your 
upload to be initiated. My question was that is the new behavior expected and 
fine?
   
   Yes, the upload will happen sooner (despite the potential wait) than it 
would have happened in the original sequential code.
   
   > I meant that the code for semaphores directly in the output stream seems 
more related to the behavior of the output stream rather than a property of the 
uploading mechanism.
   
   We don't want the main threads to be blocked by the upload threads (except 
for the last chunk where we have to complete the multipart upload). Hence, the 
main thread cannot release the semaphore post finish of upload(), if that's 
what the suggestion was.
   
   > Yes, this would only work for a single chunk.
   
   Cool, I agree then. The only counter point I have would be the 
maintainability angle. I'll discuss it with Karan offline if it makes sense to 
go for it. Either way, this can go in a separate follow-up PR, if it has to. 
Hope that works!


-- 
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