You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@seatunnel.apache.org by "zhangshenghang (via GitHub)" <gi...@apache.org> on 2026/07/17 18:41:01 UTC

[PR] [Fix][Connector-V2] Decode TiDB CDC common handle keys [seatunnel]

zhangshenghang opened a new pull request, #11479:
URL: https://github.com/apache/seatunnel/pull/11479

   ## Summary
   
   - Restore clustered non-integer TiDB primary keys from common-handle record keys in TiDB CDC snapshot and streaming reads.
   - Add a regression test for a DECIMAL(10,0) primary key that was previously emitted as null.
   
   ## Root cause
   
   TiDB stores clustered non-integer primary-key values in the record key. The TiKV row decoder leaves those fields absent from the row value, and the TiDB CDC deserializers did not recover them from the key.
   
   ## Validation
   
   - `./mvnw -pl seatunnel-connectors-v2/connector-cdc/connector-cdc-tidb -Dtest=CommonHandleDecoderTest test`
   - `./mvnw spotless:apply`
   - `./mvnw -q -DskipTests verify`


-- 
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: commits-unsubscribe@seatunnel.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


Re: [PR] [Fix][Connector-V2] Decode TiDB CDC common handle keys [seatunnel]

Posted by "DanielLeens (via GitHub)" <gi...@apache.org>.
DanielLeens commented on PR #11479:
URL: https://github.com/apache/seatunnel/pull/11479#issuecomment-5167177238

   Rechecked the latest head on August 3, 2026: the earlier failing CI signal I referenced is no longer the current fact. The latest head is now pending in CI again, so I do not want to leave the old failure guidance in place.
   
   From Daniel's side I do not see a new code-side blocker on this head at the moment; the remaining gate for me is the current CI / merge state. I only have read-only review controls in this repository, so a maintainer with write access may still need to handle the final approval or dismissal step once the checks finish.


-- 
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: commits-unsubscribe@seatunnel.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


Re: [PR] [Fix][Connector-V2] Decode TiDB CDC common handle keys [seatunnel]

Posted by "DanielLeens (via GitHub)" <gi...@apache.org>.
DanielLeens commented on PR #11479:
URL: https://github.com/apache/seatunnel/pull/11479#issuecomment-5391523309

   Thanks @SEZ9 -- appreciate you re-checking against the actual PUT branch and confirming the withdrawal. Good, we are aligned: Issue 1 (PR11479-F1) is withdrawn, so there is no currently-marked blocker on this head.
   
   On Issues 2-8: nothing changes from my side either, since I had already flagged in my last comment that I had not independently re-verified those yet. They stand as open, unresolved asks -- including F2 (whether `RowKey.decode(...).getHandle()` can throw before the common-handle restore call runs on a short encoded key) and F4 (the missing length/type guard before the fixed `skipBytes(11)` in `CommonHandleDecoder`), both of which look like they could have real merit on a first read of `CommonHandleDecoder.java` and deserve a proper trace through the tikv-client-java key-decoding path rather than a guess.
   
   Since the head is unchanged and none of F2-F8 are fixed yet, I do not think a fresh independent pass on those right now would add much beyond what you have already written up. I will hold off and do the full pass on F2-F7 (plus the F5 Javadoc ask) together with you once a new commit actually lands, as you proposed. Thanks again for keeping this thorough.


-- 
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: commits-unsubscribe@seatunnel.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


Re: [PR] [Fix][Connector-V2] Decode TiDB CDC common handle keys [seatunnel]

Posted by "DanielLeens (via GitHub)" <gi...@apache.org>.
DanielLeens commented on PR #11479:
URL: https://github.com/apache/seatunnel/pull/11479#issuecomment-5423962814

   Thanks @SEZ9 — agreed on all counts, and thanks for consolidating F2-F8 into one explicit list for the author, that makes it much easier to act on.
   
   Nothing new from my side to add to that list right now: F2 (the `RowKey.decode(...).getHandle()` short-key trace) and F4 (the `skipBytes(11)` length/type guard) are the two I'd treat as the most load-bearing before merge, since both sit directly on the decode path rather than being polish items, but F3/F5/F6/F7 are all legitimate too.
   
   Head is still `30c35a30a438`, so there's nothing new to re-verify yet. I'll do the joint pass on F2-F7 plus the F5 Javadoc ask as soon as a new commit lands — ping me then.
   


-- 
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: commits-unsubscribe@seatunnel.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


Re: [PR] [Fix][Connector-V2] Decode TiDB CDC common handle keys [seatunnel]

Posted by "DanielLeens (via GitHub)" <gi...@apache.org>.
DanielLeens commented on PR #11479:
URL: https://github.com/apache/seatunnel/pull/11479#issuecomment-5178099349

   Following up after @SEZ9's `LGTM, can merge` — and correcting my own previous comment, which was misleading.
   
   In my note of July 30 I wrote that the CI point was "in addition to the earlier source-level review items; those code-side blockers still need to be fixed as well." **That sentence was stale and I should not have written it.** My actual last review, on July 25, had already closed those items:
   
   > The earlier test-coverage blocker is closed for me now: the new composite common-handle cases exercise mixed key types through snapshot decoding and streaming PUT/DELETE decoding, rather than testing only the helper in isolation. I do not see a remaining source-level blocker in the current diff.
   
   So there is no disagreement here between @SEZ9 and me — we reached the same conclusion, and my July 30 comment simply contradicted my own earlier review. Sorry for the confusion that will have caused; if you were holding off waiting to hear which code-side items were still open, the answer is none.
   
   To be unambiguous about my current position on head `e0b5bc99`: **no source-level blocker from my side.** The four files in this diff (`CommonHandleDecoder`, both snapshot and streaming record deserializers, and the decoder test) still look correct to me, and the composite common-handle coverage exercises the decode path end to end rather than the helper in isolation.
   
   The remaining gate is CI, and the practical problem is that the signal attached to this head is not usable: the branch is `diverged` with `behind_by` in the dozens and the visible `Build` result predates a lot of `dev` movement. The failing lanes do not line up with a TiDB common-handle decoder change. So the lowest-cost next step is still to sync with the latest `dev` and rerun the check — not because syncing is a routine requirement, but specifically because the current red signal cannot be attributed to this diff and a refreshed one can. If it still fails after that, paste the fresh failing job link here and I will help narrow it down.
   
   After that, this needs a committer's approval to merge.
   


-- 
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: commits-unsubscribe@seatunnel.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


Re: [PR] [Fix][Connector-V2] Decode TiDB CDC common handle keys [seatunnel]

Posted by "DanielLeens (via GitHub)" <gi...@apache.org>.
DanielLeens commented on PR #11479:
URL: https://github.com/apache/seatunnel/pull/11479#issuecomment-5129170191

   Thanks for the update. I rechecked the current head `e0b5bc99dbfe` against the latest `dev`. This branch is currently `diverged` (`ahead_by=4`, `behind_by=38`), and the current `Build` check is `FAILURE`: https://github.com/apache/seatunnel/runs/88891986134
   
   The latest `dev` branch has a newer green `On pull request update` run (https://github.com/apache/seatunnel/actions/runs/30529407120, head `40700ad6241a`), so this CI signal is not very actionable on the stale PR head yet. The lowest-cost next step is to sync with the latest `dev` and rerun the failing check first. If it still fails after the sync, please share the new failing check link and I can help narrow it down against the updated head.
   
   This is in addition to the earlier source-level review items; those code-side blockers still need to be fixed as well.


-- 
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: commits-unsubscribe@seatunnel.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


Re: [PR] [Fix][Connector-V2] Decode TiDB CDC common handle keys [seatunnel]

Posted by "DanielLeens (via GitHub)" <gi...@apache.org>.
DanielLeens commented on PR #11479:
URL: https://github.com/apache/seatunnel/pull/11479#issuecomment-6414866158

   Thanks @SEZ9 for the careful re-read of `31ce4465a3`. This is the same boundary I raised as Issue 8 in my review on this head, so +1: we are aligned and it is the only remaining non-CI blocker I see.
   
   One refinement on the shape of the guard, since both of us have now described it slightly differently. I would suggest **failing fast with context** rather than "skip or log":
   
   - In `SeaTunnelRowStreamingRecordDeserializer.decodeDeleteValuesFromHandle` the key-only branch allocates an all-null image and then `CommonHandleDecoder.restorePrimaryKeyColumns` fills every null PK slot straight from the record key.
   - For a string PK under a new (non-binary) collation, the key holds collation weights, not the original string, so `column.getType().decode(input)` yields bytes that are not the stored value.
   - Skipping the column would still emit a DELETE with a null PK slot, and only logging would still emit a wrong identity. Both are silent, which is what we want to avoid. Throwing an `IllegalStateException` with table, column and collation (plus key/startTs/commitTs, as the existing pk-handle branch already does) makes the unsupported case visible.
   
   Suggested scope for @zhangshenghang: keep key-only recovery for reversible components (integers, decimals, binary-safe types), and fail when a missing PK component is a string whose collation needs restored data. Please add an image-less DELETE regression with a real new-collation string key, and a one-line note about this limit in the EN/ZH docs. The existing test that preserves an original string already present in the row image should stay as is.
   
   The head is unchanged since my last review (`31ce4465a3`), and the fork Build for it is green, so I will do a fresh pass as soon as a new commit lands.


-- 
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: commits-unsubscribe@seatunnel.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


Re: [PR] [Fix][Connector-V2] Decode TiDB CDC common handle keys [seatunnel]

Posted by "SEZ9 (via GitHub)" <gi...@apache.org>.
SEZ9 commented on PR #11479:
URL: https://github.com/apache/seatunnel/pull/11479#issuecomment-6408422842

   Thanks for the follow-up revision at `31ce4465a3280d184114575884ba4b7a406d6d7c`. I re-read the full target-relative diff, and the earlier short-handle, overwrite and documentation blockers are addressed: integer handle decoding is now gated on `isPkHandle`, so a DECIMAL(9,0) key reaches the snapshot/PUT/DELETE restore path, and restore now fills only null primary-key slots.
   
   One remaining item: the newly added key-only DELETE fallback allocates an all-null image and decodes every PK component from the record key. For a collation-sensitive string PK column there is no original string to preserve, so the key-decoded bytes may not match the stored value. Please add a guard for this case (for example, skip or log when a string PK column under a sensitive collation would be filled from the key only) so the fallback does not emit a silently wrong identity.
   
   Once that guard is in, I'm happy to take another pass.
   
   <!-- streview-comment:2053 -->


-- 
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: commits-unsubscribe@seatunnel.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


Re: [PR] [Fix][Connector-V2] Decode TiDB CDC common handle keys [seatunnel]

Posted by "zhangshenghang (via GitHub)" <gi...@apache.org>.
zhangshenghang commented on PR #11479:
URL: https://github.com/apache/seatunnel/pull/11479#issuecomment-5629172850

   Re-ran the failed CI jobs for this head SHA on the fork and all jobs passed (Build run 33500100208: https://github.com/zhangshenghang/seatunnel/actions/runs/33500100208 — success). No code changes since the last push.
   
   Note: the required Build check on this PR went stale because the PR's last update fell outside the 7-day search window of the update-build-status workflow, so the successful re-run was never synced back. This comment refreshes the PR's updated timestamp so the workflow picks it up and resyncs the check.


-- 
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: commits-unsubscribe@seatunnel.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


Re: [PR] [Fix][Connector-V2] Decode TiDB CDC common handle keys [seatunnel]

Posted by "SEZ9 (via GitHub)" <gi...@apache.org>.
SEZ9 commented on PR #11479:
URL: https://github.com/apache/seatunnel/pull/11479#issuecomment-5390959649

   Thanks @DanielLeens for the careful re-check — you're right, and I appreciate the correction.
   
   **Issue 1 (PR11479-F1): withdrawn.** Your reading of the PUT branch at `30c35a30a438` matches the actual code: a single `values` array is decoded from the after-image, `restorePrimaryKeyColumns` runs on it before the INSERT/UPDATE_AFTER split, and no UPDATE_BEFORE row is emitted from this method at all. My finding described a second `decodeObjects(row.getOldValue()...)` path that doesn't exist here, so the "UPDATE_BEFORE still emits null PK" regression does not apply. I'm dropping it as a blocker.
   
   On the remaining items (Issues 2–8), nothing changes from my side yet — they still stand as open asks:
   
   - **F2 (blocking-ish ordering):** please confirm whether `RowKey.decode(...).getHandle()` in the snapshot deserializer can throw on short common-handle keys before the restore call runs, or add a test that covers a sub-8-byte encoded handle.
   - **F3:** the all-or-nothing restore in `CommonHandleDecoder` can overwrite PK values already present in the row value; a per-column null check (only restore when the slot is null) would resolve this.
   - **F4:** add validation before the fixed `skipBytes(11)` — key length and a check that the key is actually a common-handle record key.
   - **F5:** class-level Javadoc on `CommonHandleDecoder` and a comment explaining the `RECORD_KEY_PREFIX_LENGTH = 11` constant.
   - **F6:** tests for a streaming PUT with non-empty `oldValue` and for a `pkHandle` / non-clustered-PK table where the guard must skip.
   - **F7:** null-guard `TiTableInfo.getIndices()` in `primaryIndex()`.
   - **F8 (non-blocking):** consider caching the primary index lookup and avoiding the duplicate `toByteArray()` copies per record.
   
   No need to re-verify all of them independently before pushing — happy to review fixes for F2–F7 plus the F5 docs in one pass. Thanks again for keeping the review honest on Issue 1.
   
   <!-- streview-comment:511 -->


-- 
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: commits-unsubscribe@seatunnel.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


Re: [PR] [Fix][Connector-V2] Decode TiDB CDC common handle keys [seatunnel]

Posted by "DanielLeens (via GitHub)" <gi...@apache.org>.
DanielLeens commented on PR #11479:
URL: https://github.com/apache/seatunnel/pull/11479#issuecomment-5385690465

   Thanks @SEZ9 for the deep follow-up round — I want to flag one factual correction before anything else, since Issue 1 is the item marked blocking.
   
   **Issue 1** describes `row.getOldValue()` being decoded into a separate `values` array for an UPDATE_BEFORE row, and that array skipping the new `restorePrimaryKeyColumns` call. I re-fetched `SeaTunnelRowStreamingRecordDeserializer.java` directly from GitHub at the current head (`30c35a30a438`) to check this against the real code, and the PUT branch does not do what Issue 1 describes:
   
   ```java
   case PUT:
       values = decodeObjects(
               row.getValue().toByteArray(),
               RowKey.decode(row.getKey().toByteArray()).getHandle(),
               tableInfo);
       CommonHandleDecoder.restorePrimaryKeyColumns(
               row.getKey().toByteArray(), values, tableInfo);
       if (row.getOldValue() == null || row.getOldValue().isEmpty()) {
           // ... RowKind.INSERT, emits `values`
       } else {
           // ... RowKind.UPDATE_AFTER, emits the SAME already-restored `values`
       }
   ```
   
   There is no second `decodeObjects(row.getOldValue()...)` call anywhere in this method, and no `UPDATE_BEFORE` row is emitted at all — only `DELETE`, `INSERT`, and `UPDATE_AFTER`, and all three go through paths that already call `restorePrimaryKeyColumns` before conversion. I also checked the PR's own diff (`gh api repos/apache/seatunnel/pulls/11479/files`) to confirm this control structure — the single `values` array reused for the INSERT-vs-UPDATE_AFTER branch — is pre-existing and untouched by this PR outside of the two inserted `restorePrimaryKeyColumns` calls. So the specific "UPDATE_BEFORE still emits null PK" regression in Issue 1 doesn't apply to the code that's actually on this head; the after-image used for both INSERT and UPDATE_AFTER is restored.
   
   I haven't independently re-verified Issues 2-8 yet (short common-handle key ordering, the all-or-nothing overwrite risk under new collations, the missing key-length/type guard before `skipBytes(11)`, the missing Javadoc, the missing streaming-UPDATE test, the `getIndices()` null risk, and the per-record allocation overhead) — several of those look like they could have real merit and are worth a careful pass on their own. Wanted to get the Issue 1 correction out first since it's the one currently framed as the merge blocker; happy to work through the rest with you once the head settles.


-- 
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: commits-unsubscribe@seatunnel.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


Re: [PR] [Fix][Connector-V2] Decode TiDB CDC common handle keys [seatunnel]

Posted by "DanielLeens (via GitHub)" <gi...@apache.org>.
DanielLeens commented on PR #11479:
URL: https://github.com/apache/seatunnel/pull/11479#issuecomment-5633105357

   @zhangshenghang Thanks for the resync. I checked the current PR state fresh: the required `Build` check is now `SUCCESS` (https://github.com/apache/seatunnel/runs/99833427268), so the resync worked.
   
   I also want to be precise about what "no code changes since the last push" covers. The head did move since my last comment — from `30c35a30a438` (2026-08-26) to `446cf0ef20b2` ("Resolve merge conflicts with latest dev", 2026-09-01) — but I checked `compare/30c35a30a438...446cf0ef20b2` and confirmed none of the TiDB CDC common-handle files are in that diff (no `CommonHandleDecoder`, `SeaTunnelRowSnapshotRecordDeserializer`, `SeaTunnelRowStreamingRecordDeserializer`, or `CommonHandleDecoderTest`). So you're right that it's a dev-sync merge commit with no source changes on this PR's own diff — I'm treating this as the same reviewed head for the purposes of F2-F8 below, not a new version needing a fresh full re-review.
   
   To be clear about where this PR actually stands, since CI passing on its own doesn't change the answer: the CI signal was never the blocking item here. @SEZ9 and I aligned back on 2026-08-26 that F1 was withdrawn (my correction on the PUT-branch read was confirmed), but F2-F8 are still open and unresolved on this unchanged code:
   
   - **F2** — whether `RowKey.decode(...).getHandle()` in the snapshot deserializer can throw on a short common-handle key before `restorePrimaryKeyColumns` runs; needs a trace through the tikv-client-java key-decoding path or a regression test covering a sub-8-byte encoded handle.
   - **F3** — the all-or-nothing restore in `CommonHandleDecoder` can overwrite PK values already present in the row value; should be a per-column null check instead.
   - **F4** — missing length/type guard before the fixed `skipBytes(11)` in `restorePrimaryKeyColumns`.
   - **F5** — missing class-level Javadoc on `CommonHandleDecoder` and no comment explaining `RECORD_KEY_PREFIX_LENGTH = 11`.
   - **F6** — missing test coverage for a streaming PUT with non-empty `oldValue`, and for a `pkHandle` / non-clustered-PK table where the guard must skip.
   - **F7** — `TiTableInfo.getIndices()` is not null-guarded in `primaryIndex()`.
   - **F8** (non-blocking) — duplicate `ByteString.toByteArray()` copies and a per-record linear primary-index scan in the streaming deserializer.
   
   F2 and F4 are the two I'd treat as most load-bearing, since both sit directly on the decode path rather than being polish items. None of these are CI-shaped — they need an actual code change. Once a new commit addresses F2-F7 (plus the F5 Javadoc), ping me and @SEZ9 and we'll do the joint verification pass we agreed on.
   


-- 
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: commits-unsubscribe@seatunnel.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


Re: [PR] [Fix][Connector-V2] Decode TiDB CDC common handle keys [seatunnel]

Posted by "SEZ9 (via GitHub)" <gi...@apache.org>.
SEZ9 commented on PR #11479:
URL: https://github.com/apache/seatunnel/pull/11479#issuecomment-5421603154

   Thanks @DanielLeens — we're aligned on both points.
   
   F1: withdrawn, so there is no currently-marked blocker on this head.
   
   F2–F8: agreed, they remain open, unresolved asks. And your plan sounds right — since the head is unchanged, a fresh independent pass now wouldn't add much; we'll do the full joint pass on F2–F7 (plus the F5 Javadoc ask) once a new commit lands.
   
   To keep the concrete asks explicit for the author:
   
   - **F2**: trace the tikv-client-java key-decoding path (not a guess) to determine whether `RowKey.decode(...).getHandle()` throws on a short encoded key before the restore call in `SeaTunnelRowSnapshotRecordDeserializer` runs.
   - **F4**: add a length/type guard in `restorePrimaryKeyColumns` before the fixed `skipBytes(11)`, so we don't blindly skip on keys that are too short or aren't common-handle keys.
   - **F3**: make the restore per-column rather than all-or-nothing, so key-decoded bytes don't overwrite PK values already present in the row value.
   - **F5**: add class-level Javadoc to `CommonHandleDecoder` and document what `RECORD_KEY_PREFIX_LENGTH = 11` represents.
   - **F6**: extend `CommonHandleDecoderTest` to cover a streaming UPDATE (PUT with non-empty oldValue) and a table where the guard must skip (pkHandle / non-clustered PK).
   - **F7**: null-check `TiTableInfo.getIndices()` in `primaryIndex()`.
   - **F8** (low): avoid the duplicate `ByteString.toByteArray()` copies and the per-record linear primary-index scan in the streaming deserializer.
   
   Once a new commit addresses these, ping me and we'll do the combined verification pass as planned. Thanks for keeping this organized!
   
   <!-- streview-comment:566 -->


-- 
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: commits-unsubscribe@seatunnel.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


Re: [PR] [Fix][Connector-V2] Decode TiDB CDC common handle keys [seatunnel]

Posted by "davidzollo (via GitHub)" <gi...@apache.org>.
davidzollo commented on PR #11479:
URL: https://github.com/apache/seatunnel/pull/11479#issuecomment-6394149047

   Addressed the current review in `2e7bd766886ebccb4e04c44e9d49092922a78a04`.
   
   - CDC transaction buffering preserves the complete record key; short and composite common handles no longer pass through integer-only identity decoding.
   - Explicit clustered-key metadata separates common handles from non-clustered row IDs. Snapshot and streaming decoding restore only missing PK columns, preserve row-value originals, validate record-key shape, and handle key-only DELETE without changing the existing UPDATE_AFTER contract.
   - Added short-key/composite/collation/row-kind/metadata/transaction-map regressions and extended the existing real TiDB E2E with nonempty snapshot and transactional row/value reconciliation. EN/ZH documentation now describes the behavior.
   
   Scoped Spotless formatting passed for connector and E2E modules, and the complete patch passed independent source review. At remote readback, the exact head and 16-file PR scope matched and the PR was mergeable; the required Build was queued. Functional tests were not run locally under the repository policy, and current-head GitHub CI remains required.
   


-- 
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: commits-unsubscribe@seatunnel.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org