SEZ9 commented on PR #11206:
URL: https://github.com/apache/seatunnel/pull/11206#issuecomment-6538944915
Thanks for the update on `d4faf1ea1aab8414aea81371c7c5112078f96739` (merge
of `dev` `3ac2d1cec6a` plus the fix commit).
**Unit test root cause / F2 and F5.** The explanation makes sense: with an
empty startup writer map and the only runtime table quarantined, the
`CONTINUE_OTHER_TABLES` all-tables-isolated rule fired inside
`applySchemaChange`, so the failure was a test-setup artifact rather than a
problem in the F2 (unsupported sink) or F5 (retry/idempotency) paths. Starting
the test from one healthy table (`db1.healthy`) and additionally asserting that
it keeps writing while no row of the rejected table reaches a writer is the
right shape. Two asks so I can close these out:
- F2: please point me to where (test or code) a runtime `CreateTableEvent`
reaching a sink that does not support newly created tables is handled
explicitly — i.e. rows for that table are either rejected with a clear error or
fail the job, never silently dropped or routed to a null writer.
- F5: please confirm the retry/idempotency contract for runtime writer
creation (what happens to binlog rows for the new table after the 3 attempts
are exhausted, and whether a later checkpoint restore re-attempts creation) and
reference the assertion that covers it.
**Remaining findings.** I have not yet seen a response on F1 (code-level
identifier enforcement for runtime table/database names in generated sink SQL
and the `${database_name}_${table_name}` template), F3
(`scan.newly-added-table.enabled` defaulting to `true` for existing wildcard
jobs), F4 (strict routing/target-scope validation for source-driven DDL), F6
(`MultiTableState` compatibility on old-checkpoint restore and rollback), F7
(JDBC sink doc update for `exactly_once = false` / `generate_sink_sql = true`),
and F8 (allowlist rejecting legal table names that start with a digit). Could
you note which of these the fix commit addresses, or reply per item with the
plan?
I'll wait for fork run 41164176211 to finish before taking another pass.
<!-- streview-comment:2174 -->
--
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]
