yujun777 opened a new pull request, #66678:
URL: https://github.com/apache/doris/pull/66678
### What problem does this PR solve?
Issue Number: CIR-21356 (internal SelectDB JIRA; no public GitHub issue yet)
Problem Summary:
Querying `information_schema.columns` could return columns attributed to the
wrong table when a table disappeared between the `getTableNames` and
`describeTables` RPCs. `describeTables` did not append a `tables_offset` entry
for the missing table, so the BE computed a wrong batch start index
(`_table_index - tables_offset.size()`) and attributed columns to the wrong
table names. The remaining IN predicate then filtered the mislabeled rows, e.g.
returning 88 rows of another table's columns instead of the expected 64.
### Release note
Fixed a correctness issue where `information_schema.columns` could return
columns attributed to the wrong table when a table was dropped concurrently
with the schema scan.
### Check List (For Author)
- Test:
- Regression test:
`regression-test/suites/information_schema_p0/test_information_schema_columns_missing_table.groovy`
(passed with the fix, failed without it)
- Full FE build with checkstyle passed
- Behavior changed: Yes (fixes wrong results of `information_schema.columns`
under concurrent DDL)
- 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]
