Gabriel39 opened a new pull request, #68993:
URL: https://github.com/apache/doris/pull/68993
### What problem does this PR solve?
External regression suites retain assumptions that no longer match native
UUID mapping, connector split planning, metadata cache ownership, and
cost-based MV selection.
- Run JDBC stale-schema setup and INSERT planning on the master FE so both
use the deliberately stale cache; retain precision-error, deduplication, and
wide-destination checks.
- Refresh the Paimon database after Spark recreates the target, and verify
it is empty before testing CTAS no-op/error behavior.
- Check Hive splitting independently of backend count while retaining
file-size, partition, and row-count assertions.
- Check successful MV candidate generation after partial Hudi refresh, where
the compensation union may cost more than a base scan; retain full-refresh
selection and unrefreshed-partition rejection checks.
- Avoid an extra path separator when constructing Azure ABFS/ABFSS URIs.
- Regenerate SQL Server schema/value expectations for native UUID columns.
### Release note
None
### Check List (For Author)
- Test:
- [x] Groovy compilation of all six modified suites.
- [x] Targeted positive/negative checks for Hive split assertions, the
existing MV candidate helper, and Azure URI construction.
- [x] SQL Server suite: reproduced the old UUID golden mismatch,
regenerated results with the regression runner, and passed the complete suite
in comparison mode (including NULLs, multi-block reads, and both JDBC drivers).
- [x] PostgreSQL stale-precision suite: passed both expected-error
checks and all post-refresh/deduplication/wide-destination checks on a
single-FE local cluster.
- Local integration tests reused existing FE/BE binaries, not a fresh
build of this branch. Multi-FE forwarding and full Hive/Hudi/Paimon/Azure
integration still require CI validation.
- Behavior changed: No production behavior changes.
- Does this need documentation: No.
--
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]
