tchivs commented on PR #4530: URL: https://github.com/apache/flink-cdc/pull/4530#issuecomment-5619084058
Two findings while waiting, one of which changes what I said earlier about needing a new JIRA. **The snapshot TIME truncation is already covered by FLINK-39748 / #4412.** `PostgresConnection#getColumnValue` deliberately reads `PgOid.TIME` and `TIMETZ` as a string, with the comment "In order to guarantee that we resolve TIMETZ columns with proper microsecond precision". The microseconds are lost only because `PostgresScanFetchTask.createDataEventsForTable` bypasses that method with a bare `rs.getObject(i + 1)` — which is exactly the bypass @JNSimba removes in #4412 for TIMESTAMP / TIMESTAMPTZ / DATE. So no separate JIRA is needed; TIME comes along for free there. That does create an ordering interaction worth flagging: once #4412 lands, `time_6_c` will arrive from the snapshot with microseconds, so the expectation I just corrected has to go back to `18:00:22.123456`. I pushed f178753, which says so in the test comment and names FLINK-39748, so whoever merges second sees why the line moves. #4412 does not currently touch `PostgresFullTypesITCase`, so if it merges after this PR its CI will flag that one line. The change-stream assertion stays valid either way. **On the remaining red job:** `Pipeline E2E Tests (4-Parallelism)` fails across unrelated branches, not just here. Of the last 20 completed `Flink CDC CI` runs: `34074617215` (#4526's branch FLINK-40572) failed the 1- and 4-parallelism E2E jobs and its next run `34308394179` was green with no code change to those tests; `33623144908` on `master` failed the 1-parallelism jobs; `34179458569` on `master` failed `Source Unit Tests 2.x`; and nightly `34072345338` failed the same 4-parallelism job on `MySqlToHudiE2eITCase#testSyncWholeDatabase` — the same `testSyncWholeDatabase` shape as the `MySqlToIcebergE2eITCase` failure here. A re-run should be enough; I still cannot trigger one myself. -- 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]
