kkhatua commented on issue #1608: DRILL-6960: AutoLimit the size of ResultSet 
for a WebUI (or REST) client
URL: https://github.com/apache/drill/pull/1608#issuecomment-461693969
 
 
   @vvysotskyi (cc @ihuzenko )
   I was discussing with Sorabh the issue of whether I can fetch only an 
initial small part of a resultset for a query and terminate it without 
consuming the remaining rows, and he pointed out that cancellation is the only 
way to do this, which is not a very graceful way of executing the query to 
fetch a limited set of rows. My current problem is that a massive join query 
was able to create the small dataset ready for the WebServer when I tried you 
approach of not building the resultset beyond what the limit was. However, I am 
still forced to iterate through the remaining incoming batches till the query 
completes. That means, while I have the benefit of not overload the WebServer 
(original intent of DRILL-6050 and DRILL-6477), the Foreman is still forced to 
run through all the remaining batches only to discard them. 
   As a result, the Calcite approach suggested by Igor seems to be the best 
approach, because it also ensures that the Foreman does not process records 
that will not be sent to the Web/REST client.
   I'm in the process of doing some final testing with the last commit on the 
Calcite-based solution, which are mostly UX related. Should I submit the commit 
ID to this PR or open a new PR for the review?

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
[email protected]


With regards,
Apache Git Services

Reply via email to