SEZ9 commented on issue #12691: URL: https://github.com/apache/seatunnel/issues/12691#issuecomment-6116784890
@xinnyuli Thanks for getting the PR up so quickly, and for keeping it to the Blob / `wasb://` scope agreed above. I'll review the linked PR itself; a few things I'd like to be able to confirm directly from it so we can keep this issue thread lightweight: 1. **Supplied-runtime docs** – please make sure the docs page enumerates the exact `hadoop-azure` 3.1.4 / Azure storage SDK jar set (artifact + version) that has been checked against a clean `lib/`, together with the license/NOTICE treatment, and states plainly that the `test`-scope POM entries do not bundle anything at runtime. 2. **IMap recovery coverage** – the core of this issue is that `HdfsWriter` relies on `hflush`/`hsync`, which WASB block blobs do not honor until `close()`. Please point me at the Azurite test that starts a new instance against an unclosed IMap WAL and proves the `AzureWriter` rewrite-per-update / `block.size` rollover path recovers the data, since that is the acceptance criterion here. 3. **Azurite isolation** – confirm the host-port requirement is fully contained in the test lifecycle, that the test fails with a clear message if the emulator endpoint is unavailable, and that no real credential or endpoint can land in test output or logs. 4. **Validation** – a quick note in the PR description on where `abfs://`/`abfss://` rejection and the missing-`azure.container` / missing-authority failures are unit-tested would help. If all four are already in place, just reply here with where to look and I'll pick it up on the PR. We'll keep this issue open until that PR lands. <!-- streview-comment:1688 --> -- 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]
