Peter Rozsa has posted comments on this change. ( 
http://gerrit.cloudera.org:8080/22459 )

Change subject: IMPALA-13611: Add interop tests for Iceberg tables
......................................................................


Patch Set 2:

(5 comments)

http://gerrit.cloudera.org:8080/#/c/22459/1/tests/stress/test_update_stress.py
File tests/stress/test_update_stress.py:

http://gerrit.cloudera.org:8080/#/c/22459/1/tests/stress/test_update_stress.py@215
PS1, Line 215:     """Increments j's value with each iteration until all rows 
are deleted"""
> This comment doesn't seem to be for this function. Neither 'num_inserts' no
Done


http://gerrit.cloudera.org:8080/#/c/22459/1/tests/stress/test_update_stress.py@226
PS1, Line 226: ve_client.close()
> Did you check if the number if deletes and updates are balanced also within
It's hard to tell what is balanced in this situation. My observation is that if 
one of the writer's clocks is not shifted a bit, they lock each other from 
committing. If I keep Hive's interval 7 times higher then Impala's deleted job 
will do multiple deletes, and Hive can do one update at least. I know it's not 
optimal, but this case will run for a lot more time without shifting.


http://gerrit.cloudera.org:8080/#/c/22459/1/tests/stress/test_update_stress.py@282
PS1, Line 282: t(
> num_rows
Done


http://gerrit.cloudera.org:8080/#/c/22459/1/tests/stress/test_update_stress.py@358
PS1, Line 358:     """Issues DELETE statements from Impala and UPDATE 
statements from Hive
> Couldn't the implementations of these two functions be merged, with the del
The whole class could be refactored to make the operators interchangeable, I 
followed the pattern that is already used here. I think, for tests, simplicity 
and readability are more important than the code size. I can rewrite it, but in 
this case, I'd refactor the whole class, and I could not promise a better 
result at the end.


http://gerrit.cloudera.org:8080/#/c/22459/1/tests/stress/test_update_stress.py@409
PS1, Line 409:
> Is it intentional that test_iceberg_impala_deletes_and_hive_updates doesn't
Yes, in this case, Hive's deletion must be processed by event processing or by 
an explicit REFRESH. In this case, I choose REFRESH. The other case will always 
show the correct result without REFRESH as the last "updater" of the table is 
Impala.



--
To view, visit http://gerrit.cloudera.org:8080/22459
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Ic2ee6d3354c3b11264c5e3ded9826831e3962a98
Gerrit-Change-Number: 22459
Gerrit-PatchSet: 2
Gerrit-Owner: Peter Rozsa <[email protected]>
Gerrit-Reviewer: Daniel Becker <[email protected]>
Gerrit-Reviewer: Gabor Kaszab <[email protected]>
Gerrit-Reviewer: Impala Public Jenkins <[email protected]>
Gerrit-Reviewer: Noemi Pap-Takacs <[email protected]>
Gerrit-Reviewer: Peter Rozsa <[email protected]>
Gerrit-Comment-Date: Mon, 24 Feb 2025 17:57:47 +0000
Gerrit-HasComments: Yes

Reply via email to