carloea2 opened a new issue, #8173:
URL: https://github.com/apache/texera/issues/8173

   ### Root cause
   
   Python tuple hashing converts timestamp keys to whole epoch seconds, while 
Scala delegates to Java `Timestamp.hashCode`, which uses epoch milliseconds. 
Equal timestamp keys can therefore choose different workers in a hash shuffle 
when Scala and Python producers feed the same downstream operator.
   
   ### Reproduction
   
   Use three receivers and the timestamp `1970-01-01T00:00:00.500Z`.
   
   Before: Python hash 31 selects worker 1, Scala hash 531 selects worker 0
   After: both hashes are 531 and both select worker 0
   
   Negative, zero, and positive millisecond boundaries are covered.
   
   ### Expected behavior
   
   Python tuple hashes match Java tuple hashes at timestamp millisecond 
precision.
   
   ### Version
   
   Current `main` at 70c21145887920528d7d5540e3fb790b43e8b759.


-- 
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]

Reply via email to