github-actions[bot] commented on code in PR #65435:
URL: https://github.com/apache/doris/pull/65435#discussion_r3556079877


##########
regression-test/suites/partition_p0/auto_partition/test_auto_new_recycle.groovy:
##########
@@ -172,14 +170,17 @@ suite("test_auto_new_recycle", "nonConcurrent") {
     sql """ admin set frontend config 
('dynamic_partition_check_interval_seconds' = '600') """
     sleep(8000)
     sql """
-        insert into auto_recycle select date_add(now(), interval number-5 day) 
from numbers("number" = "8");
+        insert into auto_recycle
+        select date_add('1900-01-01 00:00:00', interval number day) from 
numbers("number" = "5")
+        union all
+        select date_add('2900-01-01 00:00:00', interval number day) from 
numbers("number" = "3");

Review Comment:
   This removes the current-day boundary from the test. The old `now() + [-5, 
2]` data covered five historical partitions plus the partition containing the 
scheduler's current time and two future partitions, but the replacement only 
has 1900 historical partitions and 2900 far-future partitions. In 
`getDropPartitionOpForAutoPartition`, the boundary being protected is 
`upperBound.compareTo(currentTimeKey) < 0`, so a current-day partition is 
different from a far-future one: its lower bound is already before 
`currentTimeKey`, and only its upper bound keeps it out of the drop set. With 
this rewrite, a regression that classifies the current-day partition as history 
could still pass while preserving the 2900 partitions. Please keep the fixed 
far-future rows, but add deterministic coverage for the partition containing 
the scheduler's current time as well.



-- 
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