morningman commented on PR #68712:
URL: https://github.com/apache/doris/pull/68712#issuecomment-5976471807

   Addressed in ed467c58a4e and 57146a7b8c6.
   
   Per finding of the review:
   
   - M1 (merge after #68710): no code change. This PR waits for #68710, as the 
description's merge order says; details in the thread.
   - M2 (zero allocation on the TaskExecutor path): fixed in ed467c58a4e. 
`_pull_next_scan_task()` takes `expected_scanners` as it is, zero included, and 
still admits one task when nothing is occupied, as `can_admit_scan_task()` 
does. New test 
`ScannerContextTest.task_executor_keeps_zero_adaptive_allocation`, which fails 
without the change.
   - M3 (the TaskExecutor low-memory cap counts only in-flight tasks): 
pre-existing since #47462, and the same today with 
`enable_adaptive_scan=false`. It goes into a follow-up together with the two 
other places where the scan side of low-memory mode has drifted; details in the 
thread.
   
   External Regression on 8784c4bec3f ([build 
103176](http://43.132.222.7:8111/buildConfiguration/Doris_External_Regression/1062095)):
 `test_file_cache_query_limit` failed because of this PR. The per-query file 
cache limit is best-effort: once a query reaches it, a block that a running 
read still holds is admitted on top of it. At up to 16 scanners per instance 
the suite's limited query ended with 54 blocks (56.6 MB) against a 21.5 MB 
limit; it ended with 20 blocks in the runs of #68711 and #68713 on the same 
base, and the suite passed in all its other 449 runs over the past month. 
57146a7b8c6 runs the suite's scans with `max_file_scanners_concurrency=1`, as 
the condition cache suites next to it do; the description covers it under 
Results.
   


-- 
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