Gabriel39 opened a new pull request, #68819:
URL: https://github.com/apache/doris/pull/68819
### What problem does this PR solve?
Issue Number: N/A
Related PR: #68808
Cherry-pick the regression-case change from #68808 to master.
The large-string-map Parquet fixture contains two 1 GiB keys in a column
chunk larger than 2 GiB. Reading both rows into one output column requires a 4
GiB buffer allocation, which can exceed the BE process memory limit during ASAN
regression runs.
Use a statement-local `batch_size=1` hint and disable aggregate pushdown.
Both rows still undergo full Map/string decoding, including the
dictionary-to-PLAIN transition, while each output batch contains one key. The
query result remains unchanged. Oversized column-chunk coverage is retained;
accumulating both keys in a single output ColumnString is no longer exercised.
### Release note
None
### Check List (For Author)
- Test:
- [x] Manual test: Groovy 4.0.19 syntax compilation and `git diff
--cached --check` passed. Verified the picked additions/deletions exactly match
#68808 and master accepts batch_size=1.
- [ ] Regression test: HDFS/ASAN end-to-end validation was not run
locally; the original CI fixture environment is not configured.
- Behavior changed: Regression query execution settings only; production
code and expected results are unchanged.
- Does this need documentation: No.
### Check List (For Reviewer who merge this PR)
- [ ] Confirm the release note
- [ ] Confirm test cases
- [ ] Confirm document
- [ ] Add branch pick label
--
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]
