You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@seatunnel.apache.org by "goutamadwant (via GitHub)" <gi...@apache.org> on 2026/08/15 05:20:10 UTC

[PR] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

goutamadwant opened a new pull request, #11814:
URL: https://github.com/apache/seatunnel/pull/11814

   ### Purpose of this pull request
   
   Fixes #11656.
   
   This migrates `JobEventHttpReportHandler` and its focused mock-server tests from okhttp 2.x to the existing managed okhttp3 4.12.0 dependency.
   
   The existing HTTP behavior is preserved:
   
   - 30-second connection timeout
   - 10-second read and write timeouts
   - five idle connections with a five-minute keep-alive
   - configured request headers
   - existing response and error handling
   
   The okhttp3 response is closed with try-with-resources.
   
   ### Does this PR introduce _any_ user-facing change?
   
   No.
   
   The event-report configuration and HTTP request behavior remain unchanged. This is an internal dependency migration.
   
   ### How was this patch tested?
   
   Updated the focused unit coverage to verify the request method, configured headers and reported event payloads using okhttp3 MockWebServer.
   
   Verified on JDK 11 with:
   
   ```shell
   ./mvnw -pl seatunnel-engine/seatunnel-engine-server \
     -Dtest=JobEventHttpReportHandlerTest \
     -DfailIfNoTests=false test
   
   ./mvnw -pl seatunnel-engine/seatunnel-engine-server package
   
   ./mvnw -pl seatunnel-engine/seatunnel-engine-server,seatunnel-e2e/seatunnel-engine-e2e/connector-console-seatunnel-e2e \
     spotless: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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub)" <gi...@apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5658380388

   Thanks for syncing `dev` into the branch. Comparing `af713e463f3..d9855aee7d7`, the merge commit `d9855aee7` does not change the handler, its unit test, or the module POMs this PR touches; the only PR-owned files affected are the incompatible-changes docs, where the resolution just appends the unrelated `dev` entries alongside this PR's own entries. So this looks like a purely mechanical sync.
   
   That also means the previously open findings are unchanged. To close them out I still need the following on the branch (or a reason why a given one should stay as-is):
   
   - **F1 (HIGH, E2E wait strategy)** – move the `received new worker register` wait out of `executeExtraCommands` and into the container lifecycle configuration so it actually gates server readiness before the test runs.
   - **F2 (MEDIUM, close() vs. running task)** – when the scheduler does not terminate in time, `close()` should not flush the local buffer concurrently with the still-running task; please serialize the flush with the task and handle in-flight connections that `evictAll()` cannot reclaim.
   - **F3 (MEDIUM, close() exception handling)** – the blocking Hazelcast/HTTP flush in `close()` may run on an interrupted thread and currently only catches `HazelcastInstanceNotActiveException`/`IOException`; other runtime failures still escape `close()`.
   - **F4 / F6 (MEDIUM/LOW, Kotlin stdlib versions)** – compile-scope okhttp3 brings `kotlin-stdlib` 1.8.21 and `kotlin-stdlib-common` 1.9.10 onto the server classpath, and the same split shows up in the dependency and license inventories. Please either pin the Kotlin stdlib artifacts to one version or confirm the resolved dist actually ships this mix and the inventories match it.
   - **F5 (MEDIUM, `testRetryAfterHttpFailure`)** – use `takeRequest` with a timeout so the build cannot hang, close the request `Buffer`s, and don't rely on the scheduler firing immediately given the 1-day interval (trigger the flush explicitly or shorten the interval in the test).
   - **F7 (LOW, redirects)** – disable redirect following on the new `OkHttpClient` (or strip the configured headers on cross-host redirects) so report headers carrying tokens are not forwarded to another host.
   - **F8 (LOW, Content-Type)** – `RequestBody.create(String, MediaType)` appends `; charset=utf-8` when the configured media type has no charset; please confirm whether the header sent to existing collectors stays byte-identical, and if not, build the body from bytes with the original `MediaType`.
   
   If any of these were already handled in a commit I have not seen, just point me to it and I will re-check. Otherwise, once the above land I'm happy to do one more pass.
   
   <!-- streview-comment:1032 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "DanielLeens (via GitHub)" <gi...@apache.org>.
DanielLeens commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5652845546

   Thanks for the checklist — fair ask given how long this thread has run. Head is still `af713e463f`, no new commit since my last comment, so I re-read the actual current source for each item below rather than pointing back at old commit messages.
   
   **1. F1 (wait strategy in `FakeSourceToConsoleWithEventReportIT`).** I want to flag something before answering: you yourself marked this "Resolved previous findings" in your 2026-08-30 review (`f83cb22736e4`), and nothing in `FakeSourceToConsoleWithEventReportIT.java` or its parent container class has changed since then (byte-identical, per both of our confirmations this week). I re-traced the actual call chain again today rather than relying on that prior conclusion: `SeaTunnelEngineContainer.startUp()` calls `super.startUp()`, which (`SeaTunnelContainer.java:98-101`) calls `createSeaTunnelServer()` -> `createSeaTunnelServer(NETWORK)`. Inside that method, `executeExtraCommands(server)` is invoked at line 146, and `server.start()` is only called afterward, at line 148. So the custom `LogMessageWaitStrategy` this test sets inside its `executeExtraCommands` override (`FakeSourceToConsoleWithEventReportIT.java:106-117`) is applied to the container's configuration *before* `start()` r
 uns, not after — it does reliably gate readiness on this path, it's not just "happens to pass." (The ordering you're describing — `start()` before `executeExtraCommands()` — does exist elsewhere in this file, in `createSeaTunnelContainerWithFakeSourceAndInMemorySink` at `SeaTunnelContainer.java:197/199`, but that overload isn't on this test's call path; it goes through `startUp()`, not that method.) If you're seeing a different call path than this, I'd genuinely like the specific trace, since what I have here says this was correctly fixed before your 08-30 round and hasn't regressed.
   
   **2. F2/F3 (`close()` shutdown hardening).** Both fixed, confirmed against current `JobEventHttpReportHandler.java`:
   - The ringbuffer/local-buffer flush is now gated on `schedulerTerminated && !interrupted` (`:237-251`); when the scheduler doesn't terminate, or the wait was interrupted, the `else` branch just logs a warning and skips the flush entirely (`:250`) — no more concurrent flush racing an interrupted-but-still-running task.
   - The two flush calls are each wrapped in `catch (Exception e)` (`:241`, `:246`), not the narrow `HazelcastInstanceNotActiveException`/`IOException` pair.
   - `httpClient.dispatcher().cancelAll()` runs before `httpClient.connectionPool().evictAll()` in the `finally` block (`:253-254`), so in-flight connections are cancelled before eviction rather than being left parked for the keep-alive window.
   
   **3. F4/F6 (Kotlin/okio version split).** Fixed. `known-dependencies.txt` now lists `kotlin-stdlib`, `kotlin-stdlib-common`, `kotlin-stdlib-jdk7`, `kotlin-stdlib-jdk8` all at `1.9.10` (no more 1.8.21), and `seatunnel-dist/release-docs/LICENSE` matches it line for line. `seatunnel-engine-server/pom.xml` now imports `kotlin-bom:1.9.10` in `dependencyManagement` (`:33-34`), which is what pins the graph to one version instead of leaving it to Maven nearest-wins.
   
   **4. F5 (`testRetryAfterHttpFailure` hygiene).** Fixed. `JobEventHttpReportHandlerTest.java:208-209` now uses `retryServer.takeRequest(10, TimeUnit.SECONDS)` for both requests (bounded, won't hang the fork), and the two `Buffer`s are read inside try-with-resources (`:212-213`).
   
   **5. F7 (redirect following).** Fixed. `createHttpClient()` builds with `.followRedirects(false).followSslRedirects(false)` (`JobEventHttpReportHandler.java:273-274`), so a 3xx now surfaces as a non-2xx through the existing retry/log path instead of forwarding configured headers to a redirect target.
   
   **6. F8 (Content-Type charset parity).** Fixed. `testReportEvent` asserts `"application/json; charset=utf-8"` against `request.getHeader("Content-Type")` (`JobEventHttpReportHandlerTest.java:175`), so the migration's header behavior is locked in by the test, not just asserted in a review comment.
   
   All eight are in on `af713e463f`, verified against the current files just now rather than carried forward from memory. CI on this exact head is green on the two jobs that exercise this diff (`engine-v2-it` 8 and 11 on ubuntu-latest), as I reported after the last push; the only failures on that run are the `minio/minio` registry-pull issue hitting the whole CI fleet in modules this PR doesn't touch. Nothing outstanding from my side.


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "goutamadwant (via GitHub)" <gi...@apache.org>.
goutamadwant commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5748011122

   Follow-up `416e8cd9d` bounds the remaining HTTP request waits and documents why the E2E readiness override is installed before container start. All 7 HTTP handler tests pass.
   
   For the repeated requests: F1 follows the actual factory-before-start path; F2/F3 retain guarded final flushes, cancellation and interrupt restoration; F5 explicitly triggers retry and closes buffers; F7 disables both redirect modes. F4/F6: the server dependency tree and inventories resolve/list OkHttp 4.12.0 and Kotlin 1.9.10; this is not a claim about every module's test dependencies. F8: the legacy OkHttp 2.7.5 String-body overload already adds UTF-8, matching the current header assertion. Container E2E and distribution-wide validation are separate from these checks.


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "DanielLeens (via GitHub)" <gi...@apache.org>.
DanielLeens commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5805501124

   Confirmed independently, SEZ9 - I re-fetched the exact head (`ec6a7f9affc`) and checked each commit rather than trusting the description. `8f4dce57d45` ("Attribute bundled Public Suffix List data") touches only `seatunnel-dist/release-docs/LICENSE`, `NOTICE`, and the new `licenses/LICENSE-publicsuffix.txt` - no code or test file. `af9b89d9e` and `ec6a7f9affc` are both `dev`-sync merges; I listed every file each one touches and none of them is `JobEventHttpReportHandler.java`, its test, either `pom.xml`, or the console-E2E module this PR owns - all of it is unrelated dev content (amazondocumentdb, cdc-postgres, mqtt, rocketmq, redis, tablestore, fluss). So yes, no code or test changes to this PR's own files since `73f66f922`, matching what my own `APPROVED` review at this exact head already verified by direct diff.
   
   Per-item status, re-checked against source at `ec6a7f9affc`:
   
   - **F1 (E2E wait strategy):** Still configured. `executeExtraCommands` (`FakeSourceToConsoleWithEventReportIT.java:94-126`) keeps the base fixture's `LogMessageWaitStrategy.withRegEx(".*received new worker register:.*")` as the actual readiness gate (line 125); the wrapper added around it only calls `logStartupThreads()` on a failed wait to capture `jps`/`jstack` diagnostics (lines 116-123, 128-156) - it does not change what gates readiness.
   - **F2/F3 (`close()` robustness):** Unchanged from my last review. `close()` unwraps `CompletionException` before the type check, logs the expected `HazelcastInstanceNotActiveException` case at INFO without a stack trace, and everything else at ERROR with the original exception; the local-buffer flush still runs afterward regardless of which branch fired, and `cancelAll()`/`evictAll()` plus interrupt-flag restoration run unconditionally in `finally`.
   - **F4/F6 (Kotlin stdlib graph):** `seatunnel-engine-server/pom.xml` imports a scoped `kotlin-bom:1.9.10` and declares `kotlin-stdlib-jdk8` without a version. `tools/dependencies/known-dependencies.txt` and `seatunnel-dist/release-docs/LICENSE` both list exactly `kotlin-stdlib`, `-common`, `-jdk7`, `-jdk8` at `1.9.10` - I grepped both files for `1.8.21` and found nothing, so there is no remaining stale entry.
   - **F5 (`testRetryAfterHttpFailure`):** Yes. It uses `retryServer.takeRequest(10, TimeUnit.SECONDS)` (a bounded timeout, not an unbounded wait), closes both request `Buffer`s via try-with-resources, and drives the scheduler explicitly by calling `handler.report()` twice, even though the handler is constructed with `Duration.ofDays(1)` - it is not relying on the 1-day interval firing on its own.
   - **F7 (redirects):** Disabled. `createHttpClient()` sets `.followRedirects(false).followSslRedirects(false)`, and `testDoesNotFollowRedirects` covers it directly.
   - **F8 (Content-Type charset):** Byte-identical to previous behavior. `MediaType.parse("application/json")` carries no charset parameter, so the emitted header is exactly `application/json` - there is no `; charset=utf-8` suffix, and this is not one of the two intentionally documented behavior changes (only "no redirects" and "TLS 1.2/1.3 only" are).
   
   Nothing here changes my prior `Ready to merge` conclusion; the only open item remains CI evidence on this exact head, which was still not green as of my last review.


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub)" <gi...@apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-6258895061

   Thanks for the follow-up work on this. Before I can sign off I want to verify the earlier points against the actual changes at `131de7f704` rather than the summary, so it would help if you could point me to the specific diffs for each:
   
   1. **F1 – E2E wait strategy**: where the `received new worker register` wait is now applied so that it gates container startup in `FakeSourceToConsoleWithEventReportIT`, rather than being set in `executeExtraCommands`.
   2. **F2 / F3 – `close()` behaviour**: the changes that stop the local-buffer flush from running concurrently with a still-running task when the scheduler fails to terminate, cancel in-flight calls, and handle runtime failures other than `HazelcastInstanceNotActiveException`/`IOException` on an interrupted thread.
   3. **F4 / F6 – Kotlin stdlib versions**: whether `kotlin-stdlib` and `kotlin-stdlib-common` are now pinned to a single version, and that `known-dependencies.txt` and `seatunnel-dist/release-docs/LICENSE` match what actually resolves in the dist.
   4. **F5 – `testRetryAfterHttpFailure`**: the changes that add a timeout to `takeRequest()`, release the request `Buffer`s, and remove the dependency on the scheduler firing immediately given the configured interval.
   5. **F7 – redirects**: the line where redirects are disabled on the `OkHttpClient` so configured headers cannot be forwarded to another host.
   6. **F8 – Content-Type charset**: whether the reported `Content-Type` stays byte-identical for existing collectors when the configured media type has no charset, or an explicit note of the behavioural change.
   
   Once I can confirm each of these in the diff I'll finish the review.
   
   <!-- streview-comment:1867 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "DanielLeens (via GitHub)" <gi...@apache.org>.
DanielLeens commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-6337543956

   A merge-state follow-up so the earlier conclusion does not hide the current gate: GitHub now reports **merge conflicts** (`mergeable_state=dirty`) between `dev` and the unchanged head `131de7f70404cffffe8fa974fc495439afdc54c1`. The compare is `diverged`, ahead 26 and behind 48.
   
   Please sync the latest `dev` and resolve the conflicts before the next merge attempt, then obtain checks for the resulting head. This request is specifically to resolve GitHub's reported conflicts; it is not based only on the behind count. The head-specific Build remains successful; green CI does not resolve this merge conflict.
   
   The earlier source-level conclusion is unchanged; this follow-up records a new merge-state gate rather than adding another code finding on the same head. Thanks for following through on this; a write-capable maintainer will still need to perform the final approval/merge step.
   


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub)" <gi...@apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5642877340

   Thanks for the follow-up commit. The delta between `9e9dd08f9a` and `af713e463f` is limited to the `seatunnel-core/seatunnel-starter/pom.xml` cleanup and the move of `HttpReportPackagingIT.java` into the console E2E module; every file touched by the earlier findings is byte-identical to `9e9dd08f9a`, so their status carries over unchanged from the previous round rather than needing to be re-litigated here.
   
   The only remaining blocker from my side is the same as before: a CI run that completes end-to-end, including the relocated packaging test actually executing in its new module. Once that run is green I'm happy to approve.
   
   <!-- streview-comment:979 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "DanielLeens (via GitHub) via commits" <co...@seatunnel.apache.org>.
DanielLeens commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-6646266977

   Thanks for the additional review. My previous technical conclusion on the unchanged head stands, and the current checks are green. The previously discussed startup wait is installed before server.start() on this fixture path; no further source request is being added. A write-capable maintainer can reconcile the outstanding review summary and final merge gate.
   
   This is an unchanged-head discussion reply, not another complete source review. Daniel has READ permission; technical approval and GitHub's effective maintainer gate are separate.


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "DanielLeens (via GitHub)" <gi...@apache.org>.
DanielLeens commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5852841665

   Re: the "cut off" note — that comment wasn't actually truncated; it ends "...no `1.8.21` entry in either file." as a complete sentence. No repost needed either way, but wanted to clear that up in case it looked odd on your end.
   
   I did go check F5/F7/F8 directly against `ec6a7f9affc` rather than leaving this as "take your time" — here's what's actually there:
   
   **F7 (redirects).** Resolved, and more solidly than I'd given credit for. `createHttpClient()` sets both `.followRedirects(false)` and `.followSslRedirects(false)` (`JobEventHttpReportHandler.java:283-284`), so okhttp3 will never follow a redirect on this client at all — there's no header-forwarding-on-redirect concern because redirects are never followed, full stop. This is also pinned by a dedicated test, `testDoesNotFollowRedirects` (`JobEventHttpReportHandlerTest.java:237-271`): it enqueues three 307 responses with a `Location` header pointing at a second `MockWebServer`, sends one report, asserts the original server received the request with the configured `Authorization` header intact, and asserts the redirect-target server's `takeRequest(1, TimeUnit.SECONDS)` is null (line 266-267) — i.e. zero requests reached the redirect target.
   
   **F5 (retry test).** `testRetryAfterHttpFailure` (`JobEventHttpReportHandlerTest.java:206-235`) doesn't wait on the `Duration.ofDays(1)` scheduler interval at all — that duration only governs the background scheduled cadence, not manual invocation. The test calls `handler.report()` twice explicitly back-to-back (lines 220-221): the first call hits the enqueued 500 and leaves the event in the local buffer (nothing drained on failure), the second call re-sends the same buffered event and hits the enqueued 200. `takeRequest` is called with an explicit 10-second timeout both times (lines 223-224), and both response bodies are read inside a try-with-resources over `Buffer` (lines 227-229), so they're closed either way.
   
   **F8 (Content-Type).** This one is a real, verifiable difference, not just a wording question. `httpMediaType = MediaType.parse("application/json")` carries no charset, and `postEvents` calls the String-overload `RequestBody.create(events, httpMediaType)` (`JobEventHttpReportHandler.java` around the `postEvents` method). OkHttp3's static factory for that overload auto-appends a charset when the caller's `MediaType` doesn't specify one and the body is a `String` — so the header actually sent is `application/json; charset=utf-8`, not bare `application/json`. This isn't a typo or hardcoded literal in the handler; it falls out of using the String-based `RequestBody.create` factory. It's also explicitly locked in by the test at `JobEventHttpReportHandlerTest.java:190`, which asserts `"application/json; charset=utf-8"` on the recorded request. So: the header is not unchanged — any downstream collector doing exact-string matching on `Content-Type` rather than parsing it would see a d
 ifferent value than a bare `application/json`. Most JSON receivers won't care, but I'd call this worth a one-line callout to @goutamadwant to confirm it's acceptable (or trivial to pin the media type with an explicit charset if they'd rather keep the string identical), rather than something I'd block on myself.
   
   With F1/F2/F3/F4/F6 already closed in earlier rounds and F5/F7/F8 now checked directly against source with test evidence, from my side there's nothing left blocking on the code itself — the outstanding item remains the `engine-v2-it` CI run actually reaching the console E2E module on this head, as I noted in my last comment.
   


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "goutamadwant (via GitHub)" <gi...@apache.org>.
goutamadwant commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5460890656

   I traced the coordinator startup stall to the synchronous `ringbuffer.headSequence()` call in the `JobEventHttpReportHandler` constructor. This runs before the event handlers are fully loaded, so the stall happens before okhttp initialization.
   
   I moved the initial ringbuffer lookup to the scheduled reporting path and removed the three minute E2E timeout override. The first report now starts after the configured interval while preserving the existing ringbuffer sequence behavior.
   
   Added a regression test that blocks `headSequence()` and verifies handler construction still completes. All four `JobEventHttpReportHandlerTest` tests pass locally.


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub)" <gi...@apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5708002274

   Thanks for the sync, @goutamadwant. Comparing c13a5357e with 5cde0fdd8, the diff is empty for the handler, its unit test, the poms, `known-dependencies.txt`, and the two ITs; the EN/ZH `incompatible-changes.md` and `LICENSE` only pick up unrelated entries from `dev`, and the OkHttp/Okio/Kotlin license entries are unchanged from what I verified at af713e463f3/d9855aee7d7/c13a5357e. So this commit doesn't change any of the earlier findings, which is what I'd expect from a conflict-resolution merge.
   
   Remaining before I can approve:
   
   1. **CI on the new head.** You mentioned the console event-report and packaging ITs passed on the previous Java 8 run, Java 11 stopped in the preceding engine module, and the new-head run is pending. Please post back once the 5cde0fdd run finishes — I'd like to see the Java 11 leg get through the engine module and into the console E2E, since that path exercises the `received new worker register` wait strategy (PR11814-F1) and the relocated `okhttp3`/`okio` on the Zeta runtime classpath (PR11814-F4).
   2. **Kotlin stdlib versions (PR11814-F4 / PR11814-F6).** The inventories still list `kotlin-stdlib` 1.8.21 alongside `kotlin-stdlib-common` 1.9.10. Please confirm this matches what actually resolves into the dist on the new head, and let me know whether you'd prefer to align them in this PR or leave as-is with a note.
   
   The other earlier points (PR11814-F2/F3 close() behaviour, F5 test hardening, F7 redirects, F8 Content-Type charset) are unaffected by this merge; I'll re-check them on the same head once CI is green rather than ask you to re-describe them here.
   
   <!-- streview-comment:1119 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub)" <gi...@apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5788563231

   Thanks for the follow-up commits. As I read the thread, `8f4dce57d45`, `af9b89d9e` and `ec6a7f9affc` are described as license/merge-only, with no code or test changes since `73f66f922`. If that is right, the code-level items from my previous review are still open. A one-line pointer per item is enough if it was already handled; for anything still open, a small follow-up commit would be great.
   
   - **PR11814-F1 (E2E wait strategy)** – Is the `received new worker register` wait still configured in `executeExtraCommands`? If it was moved so that it actually gates server readiness, please point me to where.
   - **PR11814-F2 / PR11814-F3 (`close()` robustness)** – How does `close()` behave when the scheduler does not terminate in time? Does it still flush the local buffer while the interrupted task may be running, and does it catch anything beyond `HazelcastInstanceNotActiveException`/`IOException`? A short description of the current logic is fine.
   - **PR11814-F4 / PR11814-F6 (Kotlin stdlib graph)** – Could you point me to the diff where the Kotlin stdlib version is pinned in `seatunnel-engine-server/pom.xml`? Please also confirm that `tools/dependencies/known-dependencies.txt` and `seatunnel-dist/release-docs/LICENSE` now list a single Kotlin version matching the resolved dist (no remaining 1.8.21 entries).
   - **PR11814-F5 (testRetryAfterHttpFailure)** – Does the test now use a bounded timeout on `takeRequest()`, close the request `Buffer`s, and drive the scheduler explicitly rather than relying on it firing immediately with a 1-day interval?
   - **PR11814-F7 (redirects)** – Is redirect following disabled on the `OkHttpClient`, or is there a reason to keep the default given the configured report headers?
   - **PR11814-F8 (Content-Type charset)** – Is the emitted `Content-Type` byte-identical to the previous behaviour, or is the `; charset=utf-8` suffix one of the intentional changes documented in `incompatible-changes.md`?
   
   Nothing new from my side beyond these.
   
   <!-- streview-comment:1258 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub) via commits" <co...@seatunnel.apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-6870293426

   Thanks for the follow-up on the two low-severity items.
   
   **F7 (redirects):** The pointer to `createHttpClient` with `.followRedirects(false)` / `.followSslRedirects(false)`, plus `testDoesNotFollowRedirects` covering a 307 with an Authorization header and asserting the redirect target receives nothing, is the evidence I was after. I'll confirm it on head `06a7b0ac4f416252b4234dd4f8af48d0e9f3b414` and then resolve F7.
   
   **F8 (Content-Type charset suffix):** Understood that both the previous and the new `RequestBody.create` paths append `; charset=utf-8` for `application/json` without a charset, and that the regression assertion pins `application/json; charset=utf-8`. Since this was a source comparison rather than an old/new wire capture, I'll verify the behaviour on the same head before resolving F8. One small ask: please add a one-line comment next to that assertion noting the charset suffix is pre-existing behaviour, so nobody later "fixes" the test in the wrong direction.
   
   **Still open:** I haven't seen a response yet on F1–F6:
   - F1 – the `received new worker register` wait strategy being set from `executeExtraCommands`, after the container lifecycle is already configured.
   - F2/F3 – `close()` behaviour when the scheduler doesn't terminate (concurrent flush with the in-flight task, `evictAll()` not reclaiming in-flight connections, and blocking flush on a possibly-interrupted thread with only `HazelcastInstanceNotActiveException`/`IOException` caught).
   - F4/F6 – the mixed kotlin-stdlib 1.8.21 / kotlin-stdlib-common 1.9.10 graph in the dependency and license inventories; please confirm what the dist actually resolves to and whether you can align/pin them.
   - F5 – `testRetryAfterHttpFailure`: no-timeout `takeRequest()`, leaked request `Buffer`s, and the assumption that the scheduler fires immediately despite the 1-day interval.
   
   Either a fix or a short rationale for each is fine. Please ping me once a new head is pushed so I can do a final pass.
   
   <!-- streview-comment:2331 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub) via commits" <co...@seatunnel.apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-7024337607

   @goutamadwant thanks for retrying after the Maven Central download reset.
   
   - **Connector partition 5 on JDK 17:** this now succeeds in [job 135390227292](https://github.com/goutamadwant/seatunnel/actions/runs/44013804944/job/135390227292), so the earlier dependency-download failure is no longer blocking. Please don't retry it again.
   - **Remaining gate:** [connector-file-local-it JDK 11](https://github.com/goutamadwant/seatunnel/actions/runs/44013804944/job/135390229477) shows as cancelled, so the aggregate Build is still not a pass. Please get a completed result for that lane.
   - **Event-reporting source:** no changes are requested there for this, and the earlier review points are not being reopened. Please leave that code as it is while sorting out the lane.
   
   Syncing with dev alone is not confirmed to resolve the cancellation, because the separate LocalFile timeout follow-up is not a verified fix there. If the lane is cancelled again after a rerun, please post the job link here so we can look at it together.
   
   Maintainer approval remains separate from the CI result; I'll take another look once that lane has completed.
   
   <!-- streview-comment:2463 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "goutamadwant (via GitHub)" <gi...@apache.org>.
goutamadwant commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5691742266

   Synced dev and resolved the EN/ZH documentation conflicts in 5cde0fdd, preserving both migration notes. All seven focused HTTP handler tests passed. The actual console event-report and packaging ITs also passed on the previous Java 8 CI run; Java 11 stopped in the preceding engine module. New-head CI is pending.


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "DanielLeens (via GitHub)" <gi...@apache.org>.
DanielLeens commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5846259786

   Thanks for the follow-up. No new commit has landed since `ec6a7f9affc`, so I re-checked each of F1, F2/F3, and F4/F6 directly against the live source at that exact head rather than repeating what either of us said earlier — here's what's actually there right now.
   
   **F1.** I want to flag that this one was already closed in this thread: you wrote "I'm fine dropping the HIGH severity on that basis" back on 2026-09-19, after the `executeExtraCommands`-before-`start()` call-order trace, and I independently re-confirmed the same conclusion against this exact head on 2026-09-24. Nothing on `FakeSourceToConsoleWithEventReportIT.java` has changed since either of those comments. I just re-pulled the file again to be sure: lines 114-125 still install the `LogMessageWaitStrategy` on `.*received new worker register:.*` via `container.waitingFor(...)` inside `executeExtraCommands`, and that method is invoked before `server.start()` on this test's actual call path (`startUp()` -> `createSeaTunnelServer(NETWORK)`). So it is gating readiness correctly, and I don't think it's accurate to list this as HIGH/unaddressed again — if you're tracing a different code path than the one above, I'd genuinely like the specific line reference so I can re-check it, but 
 as things stand this was resolved two rounds ago.
   
   **F2/F3.** Also unchanged and still resolved. I re-read `close()` line by line at `ec6a7f9affc` (`JobEventHttpReportHandler.java:214-270`): the final flush only executes `if (schedulerTerminated && !interrupted)` (line 238); when the scheduler doesn't stop in time or the wait was interrupted, the `else` branch just logs a warning and skips the flush entirely (line 261) — so there's no concurrent flush racing a still-running scheduled task. Both flush calls are wrapped in `catch (Exception e)` (lines 242 and 257), and the outer `finally` unconditionally runs `dispatcher().cancelAll()`, then `connectionPool().evictAll()`, then restores the interrupt flag if needed (lines 264-268). `close()` cannot throw on this path.
   
   **F4/F6.** Also unchanged and resolved. `seatunnel-engine-server/pom.xml:33-38` imports `kotlin-bom:1.9.10` into `dependencyManagement`, and the `kotlin-stdlib-jdk8` dependency at lines 112-115 declares no version, so it resolves through that BOM. I re-grepped both inventory files fresh just now: `tools/dependencies/known-dependencies.txt` and `seatunnel-dist/release-docs/LICENSE` each list `kotlin-stdlib`, `-common`, `-jdk7`, `-jdk8` at `1.9.10` only — no `1.8.21` entry in either file.
   
   **F5/F7/F8** — sounds good, take your time confirming those against the diff directly; nothing further needed from my side there.
   
   One thing worth surfacing that's new since my last comment: the fork's `Build` run for this exact head (`ec6a7f9affc`) has since finished, and it's `failure` this time rather than still queued. The job that matters for this PR is `engine-v2-it (11, ubuntu-latest)` — I pulled the full log, and the actual failure is `CheckpointCoordinatorFailoverIT.testStreamJobFailsAfterCheckpointTriggerDispatchFailure` in module 66/67 (`connector-seatunnel-e2e-base`), which has nothing to do with HTTP event reporting or okhttp3. Because the reactor isn't run with `-fae`, it stopped there and never reached module 67 (`connector-console-seatunnel-e2e`), so `FakeSourceToConsoleWithEventReportIT` and `HttpReportPackagingIT` — this PR's own tests — didn't get exercised at all in this run. The other three failing jobs (`unit-test (11, windows-latest)`, `doris-connector-it (8, ubuntu-latest)`, `transform-v2-it-part-1 (8, ubuntu-latest)`) are in modules this PR doesn't touch either.
   
   @goutamadwant — given the source-level items are all confirmed resolved above, the one open item is getting a completed `engine-v2-it` run that actually reaches the console E2E module on this head. Could you trigger a fresh run (or push if you end up making any further change) so we get a real pass/fail on the tests this PR owns?


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub)" <gi...@apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5724665482

   Thanks for the follow-up on the CI runs and the Kotlin inventory.
   
   **F4 / F6 (Kotlin stdlib versions).** The comment above reports that at `5cde0fdd8ce8ace10b688010f968bfeeee61d6b0` both `tools/dependencies/known-dependencies.txt` and `seatunnel-dist/release-docs/LICENSE` list the four `kotlin-stdlib*` entries uniformly at `1.9.10`, with no `1.8.21` entry. If that's what is on the head, the inventory-side mismatch in F6 is resolved and I'll drop it. For F4, what remains is the runtime side: could you share the resolved kotlin-stdlib / okhttp3 versions on the Zeta server module's classpath so we can confirm it is also uniformly 1.9.10? If it isn't, please pin the version explicitly rather than relying on nearest-wins resolution.
   
   **F1 (E2E wait strategy).** Understood that `engine-v2-it (11, ubuntu-latest)` reached `connector-console-seatunnel-e2e` and passed `FakeSourceToConsoleWithEventReportIT`, while the JDK 8 leg stopped at module `66/67` on an unrelated failure and never started module 67. I'm fine not blocking on the JDK 8 gap. However, a green run doesn't address the finding itself: setting the `received new worker register` wait strategy in `executeExtraCommands` happens after the container lifecycle is configured/started, so it isn't gating server readiness — the test passes on timing rather than on the wait. Please move the wait strategy into the container configuration so it is applied as part of startup.
   
   **F2, F3, F5, F7, F8.** I haven't seen a response to these yet. Could you either push changes or comment on each?
   - F2/F3: `close()` behavior when the scheduler doesn't terminate — concurrent flush with the still-running task, `evictAll()` not reclaiming in-flight connections, and only `HazelcastInstanceNotActiveException`/`IOException` being caught.
   - F5: `takeRequest()` without a timeout in `testRetryAfterHttpFailure`, the leaked request `Buffer`s, and the implicit assumption that the scheduler fires immediately despite the 1-day interval.
   - F7: `OkHttpClient` following redirects by default with the configured report headers attached.
   - F8: whether `RequestBody.create(String, MediaType)` appending `; charset=utf-8` changes the `Content-Type` that existing collectors receive.
   
   Once F1 is moved and the F4 runtime versions are confirmed, the remaining items are small and I'd expect this to be ready.
   
   <!-- streview-comment:1140 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub)" <gi...@apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-6393258922

   Thanks for resolving the conflict with `dev`. Since the PR-owned patch at `d5328f28aa9e86afcf56cb029dbc192b2707141c` is identical to `131de7f704`, the handler points already verified in this thread still hold: redirects are rejected via `followRedirects(false)` / `followSslRedirects(false)`, `close()` only flushes once the scheduler has stopped and always runs `dispatcher().cancelAll()` and `connectionPool().evictAll()` in `finally`, and the Kotlin runtime is pinned to 1.9.10 in the module.
   
   Two things remain from my side:
   
   - **Dependency/license inventories**: the merge with `dev` left duplicate entries (OkHttp, Okio, JetBrains annotations, `kotlin-stdlib-common`, and a Public Suffix List attribution). Please dedupe so each jar/attribution appears once and the versions match what the dist actually resolves.
   - **E2E wait strategy, `testRetryAfterHttpFailure`, and the Content-Type charset**: I haven't yet seen a diff that addresses these earlier points. Could you point me to where they were handled, or let me know if you'd prefer to follow up separately?
   
   Happy to take another look after that.
   
   <!-- streview-comment:2023 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "DanielLeens (via GitHub)" <gi...@apache.org>.
DanielLeens commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-6473902535

   Thanks for the contribution. The merge-gate fact cited in my previous conclusion has changed: GitHub now reports a real merge conflict. This follow-up corrects that gate on the unchanged reviewed head; it is not another full source review or a sync request based only on being behind.
   
   Gate snapshot (2026-10-28T15:58:10+00:00), head `ee727193b9e4c46fc34dc58c14c7024b658a4c93`: [Build](https://github.com/apache/seatunnel/runs/121067566561) is `FAILURE`.
   
   The blocking jobs are [Run / unit-test (17, ubuntu-latest)](https://github.com/goutamadwant/seatunnel/actions/runs/40159375093/job/121074355248) `failure`.
   
   Observed log evidence: `Failed to execute goal org.apache.maven.plugins:maven-surefire-plugin:2.22.2:test (default-test) on project seatunnel-engine-server: There are test failures.`
   
   Please inspect or rerun only the named failed job where a rerun is appropriate. The logs establish the failing stage, but do not alone establish whether the current PR caused it. No matching upstream cure has been proved by this follow-up; a base sync alone is not a verified resolution, so no unrelated module/dependency patch is requested.
   
   My [previous source conclusion](https://github.com/apache/seatunnel/pull/11814#pullrequestreview-5643605276) remains clear for the reviewed scope. CI is not being passed by inference, and this follow-up is not an effective approval.
   
   Separately, GitHub reports merge conflicts with `dev`. Please sync the latest base and resolve the conflicts first; the resulting new head needs checks and re-review. This conflict gate is independent of CI and existing source findings.
   


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "DanielLeens (via GitHub) via commits" <co...@seatunnel.apache.org>.
DanielLeens commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-7037893266

   Thanks, @SEZ9; I checked the job manifest and agree. Connector partition 5 JDK 17 is successful in job 135390227292 and should not be retried for its old dependency error. The remaining incomplete lane is [LocalFile JDK 11](https://github.com/goutamadwant/seatunnel/actions/runs/44013804944/job/135390229477), marked cancelled; the aggregate Build is therefore not green. Please obtain a completed result for that lane and share a fresh job link if it cancels again. The event-reporting source remains clear from my previous review. No verified dev fix has been established for this cancellation, so a sync alone is not a demonstrated remedy, and no unrelated event-reporting patch is needed.
   
   Checked unchanged head `c715fe3fd690f7a2a0840b077f1895411faebd59` against `dev`: compare `diverged`, ahead 33 / behind 38. This drift is recorded independently; it does not by itself request synchronization. This is a discussion/CI follow-up, not another full source review. No local build, test or runtime execution was performed. Daniel has READ repository permission; effective maintainer approval/dismissal remains a separate repository gate. Checked 2026-11-20T11:59:57.811808+00:00.


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "goutamadwant (via GitHub)" <gi...@apache.org>.
goutamadwant commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5474305290

   @DanielLeens @SEZ9 Addressed the latest review in 89fec7c58
   
   close now waits for scheduler termination before the final flush catches wrapped shutdown failures preserves interrupt state and avoids following HTTP redirects with configured headers
   
   also added redirect shutdown retry and content type coverage the focused suite passes with 6 tests


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "DanielLeens (via GitHub)" <gi...@apache.org>.
DanielLeens commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5726052066

   Thanks for the detailed follow-up, @SEZ9 — went back through the current head (`5cde0fdd8`, unchanged since my last comment) for each item below.
   
   **F4 (Kotlin/OkHttp runtime pinning).** Checked how both are resolved on `seatunnel-engine-server`'s classpath:
   - `seatunnel-engine/seatunnel-engine-server/pom.xml` lines 30-38 import `kotlin-bom:1.9.10` into this module's `dependencyManagement`, and the `kotlin-stdlib-jdk8` dependency at lines 112-115 declares no explicit version. Maven resolves versionless dependencies against the nearest enclosing `dependencyManagement` rather than nearest-wins transitive resolution, so this is pinned to `1.9.10`, not left to chance.
   - `okhttp` is pinned the same way at the root: `pom.xml` lines 349-354 manage `com.squareup.okhttp3:okhttp` via the `${okhttp.version}` property, and the server module's own `okhttp` dependency (line 108-110) takes that managed version with no override.
   
   So both are uniformly pinned on this module's classpath by construction, not just "looks aligned in the inventory files." Happy to also paste `mvn dependency:tree -Dincludes=org.jetbrains.kotlin,com.squareup.okhttp3` output if you want the resolved tree itself rather than the pom evidence — let me know and I'll run it against this head.
   
   **F1 (wait strategy gating).** I traced the actual call order rather than assuming, and I think the timing concern doesn't apply to this test's path — happy to be corrected if I'm missing something. `FakeSourceToConsoleWithEventReportIT` reaches the server through `SeaTunnelContainer.startUp()` (no-`Network` overload) → `createSeaTunnelServer()` → `createSeaTunnelServer(NETWORK)`. In that method (`seatunnel-e2e/seatunnel-e2e-common/.../SeaTunnelContainer.java`):
   - line 160: `executeExtraCommands(server);`
   - line 162: `server.start();`
   
   `executeExtraCommands` runs *before* `server.start()` on this path, and the override in `FakeSourceToConsoleWithEventReportIT.executeExtraCommands` (lines 106-117) replaces the container's wait strategy via `container.waitingFor(...)` on that same `server` instance. Testcontainers uses whatever wait strategy is set at the time `.start()` is invoked, so the custom `received new worker register:` strategy is in place before the container actually starts and blocking `.start()` does poll it.
   
   There is a second overload, `createSeaTunnelContainerWithFakeSourceAndInMemorySink` (lines 200-244), where `server.start()` (line 232) does run before `executeExtraCommands` (line 234) — if that's the method you were looking at, the timing concern is real there, but that overload isn't in this test's call chain (`FakeSourceToConsoleWithEventReportIT` never calls it). Let me know if you were tracing a different path and I'll re-check.
   
   **F2 / F3 (close() races and narrow catch).** All three sub-issues look resolved on this head, in `JobEventHttpReportHandler.java`:
   - Concurrent flush with a still-running scheduled task: `close()` only runs the final flush `if (schedulerTerminated && !interrupted)` (lines 237-251) — if the scheduler didn't stop in time it's skipped entirely (logged at line 250), so the flush and the scheduled `report()` never touch `committedEventIndex`/`localBuffer` concurrently.
   - `evictAll()` not reclaiming in-flight connections: the `finally` block now calls `httpClient.dispatcher().cancelAll()` before `httpClient.connectionPool().evictAll()` (lines 253-254), so outstanding calls are cancelled first and their connections become evictable. This is also covered by a dedicated regression, `testInterruptedCloseCancelsInFlightRequest` (`JobEventHttpReportHandlerTest.java` lines 112-142), which asserts the in-flight call is actually canceled on `close()`.
   - Narrow catch: both flush calls in `close()` now catch `Exception` broadly (lines 241, 246), not just `HazelcastInstanceNotActiveException`/`IOException`.
   
   **F5 (retry test hygiene).** Also resolved on this head, in `JobEventHttpReportHandlerTest.testRetryAfterHttpFailure` (lines 191-220): both `takeRequest` calls now pass an explicit `10, TimeUnit.SECONDS` timeout (lines 208-209), the two `Buffer`s are opened in try-with-resources (lines 212-215), and the test drives the retry by calling `handler.report()` directly rather than relying on the scheduler, even though the handler is constructed with `Duration.ofDays(1)`.
   
   **F7 (redirects).** Confirmed already fixed — `createHttpClient()` sets `.followRedirects(false).followSslRedirects(false)` (lines 273-274), and `testDoesNotFollowRedirects` (lines 222-259) exercises it end to end with a real 307 response and asserts the redirect target never receives a request.
   
   **F8 (Content-Type charset).** This one is still genuinely open — I don't see it addressed in code or in a prior comment. `testReportEvent` (lines 174-175) actually asserts the outgoing header is `application/json; charset=utf-8`, confirming `RequestBody.create(events, httpMediaType)` does append the charset since `httpMediaType` (`MediaType.parse("application/json")`) has none set. I don't have visibility into what the legacy okhttp 2.7.5 client sent here or what downstream collectors expect, so I can't tell you myself whether this is a behavior change that matters. @goutamadwant, could you check what the pre-migration client sent as `Content-Type` and confirm whether any known collector cares about the exact header (no charset vs. `; charset=utf-8`)? If it's a real behavior change and collectors are strict about it, the fix is just constructing `httpMediaType` with the charset already applied so `RequestBody.create` doesn't need to add one.
   
   To summarize where this leaves things: F1, F2, F3, F5, F7 check out as resolved on the current head with the evidence above, F4 is pinned by construction (pom evidence above, happy to also run `dependency:tree` if you'd like it), and F8 is the one open item needing either a source dig or a code change from @goutamadwant.


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "goutamadwant (via GitHub)" <gi...@apache.org>.
goutamadwant commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5308888392

   @DanielLeens Updated close so the private okhttp connection pool is evicted from an outer finally block even if scheduler shutdown or either flush path fails.


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub)" <gi...@apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5806560939

   Thanks for the detailed per-item walkthrough against `ec6a7f9affc`.
   
   The descriptions for F1 (base fixture log-message wait remaining the readiness gate, with the wrapper only adding thread diagnostics on a failed wait), F2/F3 (`CompletionException` unwrapping, INFO/ERROR split, flush regardless of branch, unconditional `cancelAll()`/`evictAll()` and interrupt-flag restoration in `finally`), F4/F6 (`kotlin-bom:1.9.10` import, unversioned `kotlin-stdlib-jdk8`, and both inventories listing only `1.9.10`), and F5 (bounded `takeRequest(10, TimeUnit.SECONDS)`, try-with-resources on both `Buffer`s, explicit `handler.report()` calls) all line up with what I asked for. I will confirm each against the actual diff before marking them resolved rather than closing them on the summary alone. Your F5 note appears to have been cut off mid-sentence; could you finish the point about the 1-day interval so I have the full picture?
   
   I also have not yet independently checked that `8f4dce57d45`, `af9b89d9e`, and `ec6a7f9affc` leave this PR's own files untouched since `73f66f922`; I will do that as part of the same pass.
   
   Two items from my earlier review are not covered in your update:
   
   1. **F7 (redirects):** Does the `OkHttpClient` still follow redirects by default? If so, please explain why forwarding the configured report headers to a redirect target is acceptable, or disable following redirects.
   2. **F8 (Content-Type charset):** How is the request body built? Please confirm the configured media type is sent byte-identical without an appended `; charset=utf-8`, or note that the header change is intended.
   
   A short reply on F5, F7, and F8 is all I need to finish the re-review.
   
   <!-- streview-comment:1274 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub)" <gi...@apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-6245236987

   Thanks for keeping the branch current. As I understand it, head `c3b70de226d0` differs from `e72541e9fe` only through the `dev` merges (`cd05aece52`, `df38a415ec`, `c3b70de226`), with no change to the PR's own lines, so my earlier points still stand until I can verify them against the diff. Could you confirm or point me to the relevant changes for each?
   
   - **Redirects**: is `followRedirects(false)` / `followSslRedirects(false)` set on the new client so configured report headers are not forwarded to another host?
   - **`close()` robustness**: does `close()` flush only after the scheduler has actually terminated and the thread is not interrupted, catch and log all flush failures (not just `HazelcastInstanceNotActiveException`/`IOException`), and cancel the dispatcher, evict the pool and restore the interrupt flag in `finally`?
   - **Kotlin graph**: is the Kotlin stdlib pinned to a single version (e.g. via a BOM), with the known-dependencies list and LICENSE inventory carrying only that version?
   - **E2E wait strategy**: the `received new worker register` wait strategy is still set in `executeExtraCommands`, after the container lifecycle is configured, so it does not reliably gate server readiness. Could you move it onto the container lifecycle?
   - **Retry test**: please use the timeout variant of `takeRequest()`, close the request `Buffer`s, and either trigger the flush explicitly or use a short interval rather than relying on the 1-day scheduler firing immediately.
   - **Content-Type**: okhttp3 `RequestBody.create(String, MediaType)` appends `; charset=utf-8` when the configured media type has no charset. Could you confirm, ideally with a test assertion, that the reported header stays byte-identical for existing collectors?
   
   Happy to re-review once these are in or confirmed.
   
   <!-- streview-comment:1840 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "goutamadwant (via GitHub) via commits" <co...@seatunnel.apache.org>.
goutamadwant commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-6651162546

   @SEZ9 Checked each item against the current head (line numbers at that SHA):
   
   - **F1**: `createSeaTunnelServer` calls `executeExtraCommands(server)` at `SeaTunnelContainer.java:162`, then `server.start()` at `:164`. So the wait set at `FakeSourceToConsoleWithEventReportIT.java:114-125` is installed before start. The config copied in the same hook (`:108-110`) also depends on that order, and the test receives events.
   - **F2**: the final flush runs only if the scheduler terminated and the thread wasn't interrupted (`JobEventHttpReportHandler.java:238`). `finally` cancels in-flight calls before `evictAll()` (`:264-265`).
   - **F3**: both flushes catch `Exception` (`:242`, `:257`) and the interrupt flag is restored (`:266-267`). Covered by tests at `JobEventHttpReportHandlerTest.java:127` and `:398`.
   - **F4**: `seatunnel-engine-server/pom.xml:30-41` imports `kotlin-bom` 1.9.10.
   - **F5**: the test calls `report()` explicitly (`:220-221`), bounds `takeRequest` at 10s (`:223-224`) and closes the buffers (`:227-228`).
   - **F6**: `known-dependencies.txt` lists only 1.9.10 and the dependency check passes. The Kubernetes block in `release-docs/LICENSE` still said 1.8.21; fixed in `fbd73878d6`.
   - **F7**: redirects are disabled at `JobEventHttpReportHandler.java:284-285`, tested at test `:238` and documented in `event-listener.md`.
   - **F8**: the okhttp 2.7.5 overload also appended `; charset=utf-8`, so the header is unchanged. `testReportEvent` asserts it (`:190`).
   
   Is anything still blocking?
   


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "DanielLeens (via GitHub)" <gi...@apache.org>.
DanielLeens commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5645143313

   Thanks, @SEZ9 — confirming your read of the delta and updating the CI picture with what's landed since my last review.
   
   **Delta confirmation.** Agreed: `af713e463f` vs `9e9dd08f9a` is exactly the two files I traced in my last review — the `seatunnel-starter/pom.xml` test-scope dependency cleanup and the `HttpReportPackagingIT.java` relocation into `connector-console-seatunnel-e2e`. Every other file in this PR is byte-identical to what's already been re-confirmed across the last several rounds, so there's nothing to re-litigate there.
   
   **CI update — the run has since completed, and the piece we were both waiting on is green.** At the time of my last review, `engine-v2-it (11, ubuntu-latest)` was still `queued`. It's since finished, along with the rest of the matrix. Pulling the fork run (`goutamadwant/seatunnel` run `34619448739`, this exact head) directly:
   
   - `Run / engine-v2-it (8, ubuntu-latest)`: `success`
   - `Run / engine-v2-it (11, ubuntu-latest)`: `success`
   
   Those are the two jobs that actually exercise this PR's diff — `FakeSourceToConsoleWithEventReportIT` and the relocated `HttpReportPackagingIT`, on both JDK 8 and JDK 11. That resolves the one blocker from my last review; I have no outstanding source-level or CI-completion concern on this head.
   
   **On the overall run showing `failure`:** I checked why. Eight jobs failed — `all-connectors-it-2` (8 and 11), `all-connectors-it-6` (8 and 11), `all-connectors-it-7` (8 and 11), and `paimon-connector-it` (8 and 11) — and every single one fails on the identical cause: `DatabendIT`, `IcebergSourceIT`, and `PaimonWithS3IT` all throw `ContainerFetchException`/`NotFoundException: pull access denied for minio/minio, repository does not exist or may require 'docker login'` trying to pull `minio/minio:RELEASE.2024-06-13T22-53-53Z`. None of those modules appear anywhere in this PR's `git diff --stat` (`seatunnel-engine-server`, `seatunnel-starter`, `connector-console-seatunnel-e2e`, docs, LICENSE, dependency manifest). This also isn't isolated to this PR — I'm seeing the same `minio/minio` pull-access-denied failure tonight on an unrelated PR's fork run as well, which points to a registry-side issue (Docker Hub now denying anonymous/unauthenticated pulls of that pinned tag) affectin
 g the whole CI fleet right now, not anything in this diff.
   
   Given both jobs that matter for this PR are green and the eight failures are conclusively unrelated (different modules, same registry-access error reproducing across PRs tonight), I'd treat this head as ready from my side — happy to see it move to approval rather than waiting on a rerun of the unrelated S3-backed connector jobs.
   


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "goutamadwant (via GitHub)" <gi...@apache.org>.
goutamadwant commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5464064563

   @DanielLeens addressed the latest review items
   
   - replaced the test import with SeaTunnel's `CompletableFuture`
   - separated initial cursor setup from actual ringbuffer overflow recovery so the first report no longer logs a false warning
   - restored the scheduler initial delay to `0` to preserve the existing first report timing
   - added regression coverage for initial cursor setup and later overflow recovery
   
   Verified the focused handler tests and the exact import policy test. I also ran the complete `seatunnel-engine-server` package locally with 383 tests and no failures
   
   The fixes are pushed in `f83cb2273` and fresh CI is queued


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "DanielLeens (via GitHub) via commits" <co...@seatunnel.apache.org>.
DanielLeens commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-6557124260

   Thanks for keeping this contribution moving. I need to update the merge-gate information on unchanged head `1e406dee272e17471241bfa43eda786ab48eb738`: GitHub currently reports actual content conflicts with `dev` (`CONFLICTING` / `DIRTY`). This supersedes the earlier no-conflict/blocked snapshot; it is a content-conflict request, not a sync request based solely on branch age.
   
   Please sync the latest `dev` and resolve the conflicts while preserving both this PR's intended change and the intervening upstream changes, then obtain required checks on the resulting head.
   
   My latest source conclusion remains clear. The separately diagnosed integration-lane timeout remains an execution gate; do not change reporter semantics just to chase that timeout. The [preceding complete source review](https://github.com/apache/seatunnel/pull/11814#pullrequestreview-5721025746) remains the source assessment; this is a gate clarification, not another full review or a new list of source findings.
   
   Current existing-head checks: [Build](https://github.com/apache/seatunnel/runs/123937253173)=TIMED_OUT. Compare `diverged`, ahead 29 / behind 9. Checked 2026-11-01T11:08:50.565180+00:00. Daniel has READ repository permission; final effective approval/dismissal and merge belong to a write-capable maintainer. Happy to review the resulting head once the conflict is resolved.
   


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub) via commits" <co...@seatunnel.apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-6567804078

   Quick status update on this PR.
   
   GitHub currently reports content conflicts with `dev` on the current head `1e406dee272e17471241bfa43eda786ab48eb738` (`CONFLICTING` / `DIRTY`), and the [Build](https://github.com/apache/seatunnel/runs/123937253173) run on that head is `TIMED_OUT`. Could you rebase onto the latest `dev`, resolve the conflicts while keeping both this PR's intended change and the intervening upstream changes, and push so the required checks can run on the new head? Please don't change the reporter semantics just to chase the integration-lane timeout — that is a separate execution gate.
   
   The [earlier source review](https://github.com/apache/seatunnel/pull/11814#pullrequestreview-5721025746) still stands as the source assessment, so no new findings here. The items from that review that are still open and that I'd like addressed (or explicitly answered) on the resulting head:
   
   - **Test (HIGH)** – `FakeSourceToConsoleWithEventReportIT`: the wait strategy on `received new worker register` is set in `executeExtraCommands`, after the container lifecycle is already configured/started, so it doesn't reliably gate server readiness.
   - **Robustness** – in the HTTP report handler's close path: when the scheduler fails to terminate, the local buffer is still flushed concurrently with the interrupted-but-running task, and `evictAll()` cannot reclaim in-flight connections; the blocking Hazelcast/HTTP flush also runs on a possibly-interrupted thread and only `HazelcastInstanceNotActiveException`/`IOException` are caught, so other runtime failures escape.
   - **Compatibility / Docs** – compile-scope okhttp3 pulls a mixed Kotlin stdlib graph (kotlin-stdlib 1.8.21 vs kotlin-stdlib-common 1.9.10) onto the Zeta server classpath without version pinning, and the license/dependency inventories show the same split. Please confirm they match the actual resolved dist and consider aligning the versions.
   - **Test** – `testRetryAfterHttpFailure` uses no-timeout `takeRequest()` (the build can hang), leaks the request Buffers, and assumes the scheduler fires immediately despite a 1-day interval. Please bound the wait, close the buffers, and trigger the flush deterministically.
   - **Security (LOW)** – the new `OkHttpClient` follows redirects by default, so configured report headers (often API tokens) can be forwarded to a redirect target on another host. Consider disabling redirect following.
   - **Functional (LOW)** – `RequestBody.create(String, MediaType)` appends `; charset=utf-8` to the Content-Type when the configured media type has no charset. Please confirm the header sent to existing collectors stays byte-identical, or document the change.
   
   Once the rebase is pushed and checks are green, I'll take another look at the resulting head. Thanks for the continued work on this!
   
   <!-- streview-comment:2209 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub)" <gi...@apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5691564349

   Thanks @goutamadwant. Confirmed `c13a5357e` is a pure sync: diffing `d9855aee7d7..c13a5357e` scoped to the 11 PR-owned files shows no change, the branch has no divergence from `dev` (`0d9f9e2303`) beyond its own commits, and the "HTTP redirects are not followed" entry in both incompatible-changes docs is intact and not duplicated. Redirect handling (F7) is therefore addressed in both code and docs.
   
   Because the merge touched none of the reviewed files, the other findings from the `d9855aee7d7` review stand exactly as written, and I'll close each from the diff rather than from a description — no checklist needed. Two clarifications:
   
   - **F1**: this is structural — the wait strategy on "received new worker register" is set in `executeExtraCommands`, after the container lifecycle is configured, so a green engine E2E run would not show that it gates server readiness. It needs a code change, not a test result. That said, please still post the broader engine E2E outcome here once it finishes.
   - **F2 / F3**: the `close()` path was hardened in this PR; these are the residual points on that hardened path (flush racing a still-running task when the scheduler fails to terminate, and runtime failures other than `HazelcastInstanceNotActiveException`/`IOException` escaping `close()`). F4/F6 (kotlin-stdlib 1.8.21 vs kotlin-stdlib-common 1.9.10), F5, and F8 are likewise unchanged.
   
   Once the E2E result and the follow-up commits land, I'll do a final pass.
   
   <!-- streview-comment:1082 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "DanielLeens (via GitHub) via commits" <co...@seatunnel.apache.org>.
DanielLeens commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-6850575194

   Thanks @SEZ9 for naming F7 and F8 explicitly. I checked the current head `06a7b0ac4f416252b4234dd4f8af48d0e9f3b414` and the two RequestBody implementations to answer those points:
   
   - **F7:** Redirects are explicitly disabled in [createHttpClient](https://github.com/apache/seatunnel/blob/06a7b0ac4f416252b4234dd4f8af48d0e9f3b414/seatunnel-engine/seatunnel-engine-server/src/main/java/org/apache/seatunnel/engine/server/event/JobEventHttpReportHandler.java#L272): `.followRedirects(false)` and `.followSslRedirects(false)` at lines 284-285. The existing [testDoesNotFollowRedirects](https://github.com/apache/seatunnel/blob/06a7b0ac4f416252b4234dd4f8af48d0e9f3b414/seatunnel-engine/seatunnel-engine-server/src/test/java/org/apache/seatunnel/engine/server/event/JobEventHttpReportHandlerTest.java#L238) supplies a 307 Location and an Authorization header, verifies that the original endpoint received the request, and asserts that the redirect target receives none. This is present code/test evidence, not a new local test run.
   - **F8:** For the handler's `MediaType.parse("application/json")`, both OkHttp 2.7.5 `RequestBody.create(MediaType, String)` (RequestBody.java:46-56) and OkHttp 4.12.0 `String.toRequestBody`, exposed to Java as `create(String, MediaType)` (RequestBody.kt:105-118), append `; charset=utf-8` when the media type has no charset and encode with UTF-8. Thus this change does not newly introduce that charset suffix. The [current regression assertion](https://github.com/apache/seatunnel/blob/06a7b0ac4f416252b4234dd4f8af48d0e9f3b414/seatunnel-engine/seatunnel-engine-server/src/test/java/org/apache/seatunnel/engine/server/event/JobEventHttpReportHandlerTest.java#L189) expects exactly `application/json; charset=utf-8`. I compared the dependency source implementations; I did not execute an old/new wire-capture test.
   
   These pointers address the two outstanding questions without asking the author to duplicate code already present. Please retain your own verification step for F1-F8; I am not resolving your review threads on your behalf. My technical approval and the green exact-head CI remain separate from the required maintainer approval/dismissal gate.
   


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "goutamadwant (via GitHub) via commits" <co...@seatunnel.apache.org>.
goutamadwant commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-7003514338

   I retried Java 17 connector shard 5 after Maven Central reset the dependency download. [Retry job](https://github.com/goutamadwant/seatunnel/actions/runs/44013804944/job/135390227292)


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub)" <gi...@apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5673898343

   @goutamadwant thanks for the sync in c13a5357e and for running the focused HTTP handler tests on Java 11.
   
   The per-finding walkthrough posted earlier in this thread against d9855aee7d7 already gives concrete pointers for all of F1–F8: the wait strategy being installed before the server starts, the `close()` flushes gated on a terminated/non-interrupted scheduler with broadened exception handling and cancel-before-evict, the Kotlin BOM pin with aligned inventories, bounded `takeRequest` calls, redirects disabled, and the `Content-Type` assertion of `application/json; charset=utf-8`. I'll re-verify each of those directly against the diff in c13a5357e rather than ask you to restate them.
   
   The one item still open from my side is F1: since the wait-strategy ordering is exercised by the engine E2E run, please ping once the pending run finishes so I can confirm the event-report console IT passes there. Once that's green and my pass over the hunks checks out, I'm happy to move this forward.
   
   <!-- streview-comment:1056 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "goutamadwant (via GitHub)" <gi...@apache.org>.
goutamadwant commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5302986795

   @DanielLeens Checked the values against okhttp 2.7.5
   
   The previous client used 30 seconds for connect and the okhttp defaults of 10 seconds for read and write with a 5 connection 5 minute pool. The okhttp3 configuration keeps the same behavior and I added a short comment for this
   
   JobEventHttpReportHandlerTest passes locally with 2 tests


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "goutamadwant (via GitHub)" <gi...@apache.org>.
goutamadwant commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5378192110

   @DanielLeens I checked all four fork attempts. The remaining failure happens before the event-reporting request path. FakeSourceToConsoleWithEventReportIT times out waiting for worker registration while the server remains in coordinator initialization.
   
   Attempts 2, 3, and 4 have the same startup signature, with no HTTP assertion, classloading, or request failure. The focused handler tests are green. Since this is still an E2E test touched by the PR, I do not want to call it unrelated or retry blindly. Is this coordinator startup timeout a known CI issue, or is there a preferred environment for confirming this E2E?


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "goutamadwant (via GitHub)" <gi...@apache.org>.
goutamadwant commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5772194308

   The dev conflict is resolved while retaining both migration notes; the license, redirect and cleanup changes remain intact. I traced the failed lanes to the logical-checkpoint metrics fixture, Windows PayPal fixture, worker-loss cancellation, backpressure checkpoint progress, and Flink 1.15.3 FilterRowKind row-count fixture. Existing candidate fixes cover several of these paths; I have not weakened assertions or increased reporting timeouts to make the build green.
   
   Pushed in ec6a7f9affcf. CI for the new head is pending; the results above are scoped verification, not a green aggregate-build claim.


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub)" <gi...@apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5650998830

   Thanks for the update on the delta and the CI picture.
   
   On the `af713e463f` vs `9e9dd08f9a` delta: agreed that it is limited to the `seatunnel-starter/pom.xml` test-scope cleanup and the `HttpReportPackagingIT.java` relocation into `connector-console-seatunnel-e2e`. I also take the point that the `minio/minio` pull-access-denied failures on run `34619448739` are in modules this PR does not touch, and that `engine-v2-it (8, ubuntu-latest)` / `engine-v2-it (11, ubuntu-latest)` are reported green there.
   
   That said, a green `engine-v2-it` does not by itself close out the earlier review items, since none of them were about the packaging test or CI completion. Before I can approve, could you confirm the status of each of these on the current head:
   
   1. **PR11814-F1 (HIGH, Test)** – `FakeSourceToConsoleWithEventReportIT`: the wait strategy on `received new worker register` is still set in `executeExtraCommands`, after the container lifecycle is configured, so it does not reliably gate server readiness. A passing run does not demonstrate this is fixed; please move the wait into the container configuration (or explain why the current placement is sufficient).
   2. **PR11814-F2 / F3 (MEDIUM, Robustness)** – `close()` in `JobEventHttpReportHandler.java`: (a) when the scheduler does not terminate, the local buffer is still flushed concurrently with the interrupted-but-running task and `evictAll()` cannot reclaim in-flight connections; (b) the blocking Hazelcast/HTTP flush runs on a possibly-interrupted thread and only `HazelcastInstanceNotActiveException`/`IOException` are caught. Has either of these changed?
   3. **PR11814-F4 / F6 (MEDIUM/LOW, Compatibility & Docs)** – compile-scope okhttp3 pulling a mixed Kotlin stdlib graph (`kotlin-stdlib` 1.8.21 vs `kotlin-stdlib-common` 1.9.10) onto the Zeta server runtime classpath without pinning, and the same split in `tools/dependencies/known-dependencies.txt` and `seatunnel-dist/release-docs/LICENSE`. Please confirm the inventories match the actual resolved dist and consider aligning the versions.
   4. **PR11814-F5 (MEDIUM, Test)** – `testRetryAfterHttpFailure`: no-timeout `takeRequest()` (possible build hang), leaked request `Buffer`s, and the implicit assumption that the scheduler fires immediately despite the 1-day interval.
   5. **PR11814-F7 (LOW, Security)** – the new `OkHttpClient` follows redirects by default, so configured report headers can be forwarded to a redirect target on another host. Please consider disabling redirect following or documenting the behavior.
   6. **PR11814-F8 (LOW, Functional)** – `RequestBody.create(String, MediaType)` appends `; charset=utf-8` when the configured media type has no charset; please confirm the emitted `Content-Type` remains byte-identical for existing collectors.
   
   If any of these were already addressed in an earlier commit, a short pointer to where is enough. Otherwise a follow-up push covering F1, F2/F3 and F5 at minimum would let me move this forward.
   
   <!-- streview-comment:1017 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub)" <gi...@apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5825773436

   Thanks for the detailed follow-up on `ec6a7f9affc`.
   
   **F8** — The explanation makes sense: if the okhttp3 `RequestBody.create(String, MediaType)` overload adds `; charset=utf-8` when the `MediaType` has none, and the pre-PR okhttp 2.x call did the same, then the header existing collectors see is preserved rather than changed by this migration. I haven't verified that against the diff myself, so I'll treat this as pending until I can confirm it in the code.
   
   **F5 / F7** — Likewise, the described changes (explicit `handler.report()` calls instead of relying on the `Duration.ofDays(1)` interval, `takeRequest` bounded at `10, TimeUnit.SECONDS`, request `Buffer`s closed via try-with-resources; `.followRedirects(false).followSslRedirects(false)` covered by `testDoesNotFollowRedirects` with a 307) would address the findings. I'd like to confirm them against the diff before closing them out.
   
   I don't see updates on the remaining points yet — could you point me to the change or share your plan?
   
   1. **F1 (HIGH)** — The E2E wait on `received new worker register` is applied in `executeExtraCommands`, after the container lifecycle is already configured/started, so it doesn't reliably gate readiness. Please move it into the container wait-strategy setup.
   2. **F2 (MEDIUM)** — When the scheduler doesn't terminate in time, `close()` still flushes the local buffer while the interrupted task may still be running, and `evictAll()` can't reclaim in-flight connections. Please guard the final flush on successful termination or otherwise serialize it with the running task.
   3. **F3 (MEDIUM)** — `close()` performs blocking Hazelcast/HTTP work on a possibly-interrupted thread and only catches `HazelcastInstanceNotActiveException`/`IOException`, so other runtime failures can escape. Please broaden the handling so `close()` never throws.
   4. **F4 (MEDIUM) / F6 (LOW)** — Compile-scope okhttp3 brings a split Kotlin stdlib graph (kotlin-stdlib 1.8.21 vs kotlin-stdlib-common 1.9.10) onto the Zeta server classpath, and the same split appears in the dependency and license inventories. Please pin the Kotlin stdlib to a single version, or confirm the resolved dist artifacts and align the inventories to match.
   
   Happy to take another look once these are addressed.
   
   <!-- streview-comment:1298 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "goutamadwant (via GitHub) via commits" <co...@seatunnel.apache.org>.
goutamadwant commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-6940439167

   Synced with dev and kept the OkHttp/Okio relocations alongside the Hazelcast resource filter. The cancellation/close guards and redirect restrictions are already in this branch. Kotlin remains aligned at 1.9.10, and the Content-Type assertion now explains why no charset is appended.


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "DanielLeens (via GitHub) via commits" <co...@seatunnel.apache.org>.
DanielLeens commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-7007470675

   Thanks for the rerun. I verified that connector partition 5 JDK 17 now succeeds on this head ([job 135390227292](https://github.com/goutamadwant/seatunnel/actions/runs/44013804944/job/135390227292)); the earlier Maven-download failure in that lane is no longer the remaining gate. The latest jobs list shows only [connector-file-local-it JDK 11](https://github.com/goutamadwant/seatunnel/actions/runs/44013804944/job/135390229477) cancelled, so the aggregate Build is still not a pass. Please obtain a completed result for that lane rather than changing the event-reporting source or retrying partition 5 again. My prior source-clear conclusion remains; the independent LocalFile timeout follow-up is not a verified fix in current dev, so I am not claiming that syncing alone resolves this cancellation. Effective maintainer approval remains separate.


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub)" <gi...@apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5882640017

   Thanks for keeping the branch in sync with `dev`. From what I can see, `ec6a7f9affc...e72541e9fe18` is a pure `dev` merge that doesn't touch this PR's own files, so it doesn't change the earlier review points.
   
   Before merging I'd like to close those out explicitly on this head. Could you confirm (or point me to where it was handled) for each:
   
   - **F1 (E2E readiness wait)** — is the `received new worker register` wait strategy now applied when the container is configured rather than in `executeExtraCommands`, so it actually gates server readiness in `FakeSourceToConsoleWithEventReportIT`?
   - **F2 / F3 (`close()` robustness)** — when the scheduler fails to terminate, does `close()` still flush the local buffer concurrently with the interrupted task, and can failures other than `HazelcastInstanceNotActiveException`/`IOException` still escape `close()`?
   - **F4 / F6 (Kotlin stdlib split)** — are we still resolving `kotlin-stdlib` 1.8.21 alongside `kotlin-stdlib-common` 1.9.10 in the dist, and do `known-dependencies.txt` and the release LICENSE match what actually resolves? Pinning to a single version would be preferable.
   - **F5 (`testRetryAfterHttpFailure`)** — does `takeRequest()` now have a timeout, are the request `Buffer`s closed, and does the test no longer rely on the scheduler firing immediately despite the 1-day interval?
   - **F7 (redirects)** — is redirect following disabled on the `OkHttpClient` so configured headers/tokens can't be forwarded to another host?
   - **F8 (Content-Type)** — have you verified the `; charset=utf-8` suffix added by `RequestBody.create(String, MediaType)` doesn't change the header existing collectors see, or switched to a form that preserves the configured media type exactly?
   
   A short per-item status (fixed in which commit, or intentionally left as-is with rationale) is all I need; happy to do a final pass once that's up.
   
   <!-- streview-comment:1392 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "goutamadwant (via GitHub)" <gi...@apache.org>.
goutamadwant commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5503591690

   @DanielLeens Synced the branch with current dev and reran the focused event reporter tests
   
   JobEventHttpReportHandlerTest passes all 6 cases including retry redirect handling deferred ringbuffer initialization and shutdown behavior
   
   CI is now running on the updated head


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub) via commits" <co...@seatunnel.apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-6621276374

   Thanks for pushing this forward — good to see CI green on head `574571ff244847159796a48cef98f0756547d8a3` (apache-side Build `124986677938` is `SUCCESS` and fork run `41352860976` completed).
   
   A green build doesn't by itself close out the points from my earlier review, so before merging I'd like to get these resolved or hear why they aren't needed:
   
   **Blocking**
   - **PR11814-F1 (E2E wait strategy)** — In `FakeSourceToConsoleWithEventReportIT`, the wait on `received new worker register` is set inside `executeExtraCommands`, i.e. after the container lifecycle has already been configured/started, so it does not reliably gate server readiness. Please move the wait strategy to where the container is configured, or explain why the current placement is sufficient.
   
   **Should fix before merge**
   - **PR11814-F2 / F3 (`JobEventHttpReportHandler` `close()`)** — When the scheduler fails to terminate, `close()` still flushes the local buffer while the interrupted task may still be running, and `evictAll()` cannot reclaim in-flight connections. The flush is also a blocking Hazelcast/HTTP call on a possibly-interrupted thread, and only `HazelcastInstanceNotActiveException`/`IOException` are caught, so other runtime failures escape `close()`. Please guard the flush on successful termination (or otherwise serialize it with the task) and broaden the exception handling so `close()` cannot throw.
   - **PR11814-F4 / F6 (Kotlin stdlib graph)** — The compile-scope okhttp3 dependency brings `kotlin-stdlib 1.8.21` alongside `kotlin-stdlib-common 1.9.10` onto the Zeta server runtime classpath without version pinning, and the split is mirrored in `known-dependencies.txt` and the release-docs LICENSE. Please pin the Kotlin stdlib artifacts to a single version and confirm the inventories match what the dist actually resolves.
   - **PR11814-F5 (`testRetryAfterHttpFailure`)** — Please use a timeout when calling `takeRequest()` so a regression can't hang the build, close the request Buffers, and make the trigger explicit rather than relying on the scheduler firing immediately despite the 1-day interval.
   
   **Please confirm (low)**
   - **PR11814-F7** — The new `OkHttpClient` follows redirects by default, so configured report headers (often API tokens) could be forwarded to a redirect target on another host. Consider disabling redirect following, or confirm this is acceptable.
   - **PR11814-F8** — okhttp3 `RequestBody.create(String, MediaType)` appends `; charset=utf-8` to the `Content-Type` when the configured media type has no charset. Please confirm the reported header stays byte-identical for existing collectors, or build the body from bytes to keep it unchanged.
   
   Once F1 is fixed and F2–F5 are addressed (or explained), I'm happy to do a final pass and merge.
   
   <!-- streview-comment:2260 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub) via commits" <co...@seatunnel.apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-6538000365

   Thanks for the work so far. A quick status update for head `ee727193b9e4c46fc34dc58c14c7024b658a4c93`:
   
   **CI**
   [Build](https://github.com/apache/seatunnel/runs/121067566561) is `FAILURE`, with [Run / unit-test (17, ubuntu-latest)](https://github.com/goutamadwant/seatunnel/actions/runs/40159375093/job/121074355248) failing on `seatunnel-engine-server` with surefire test failures. Since `JobEventHttpReportHandlerTest` lives in that module, could you check the surefire report for that job and share which test(s) failed? If `testRetryAfterHttpFailure` is among them, note that PR11814-F5 already flagged its no-timeout `takeRequest()` and its assumption that the scheduler fires immediately despite the 1-day interval; adding a timeout, closing the request `Buffer`s, and triggering the flush deterministically should make it stable.
   
   **Merge conflict**
   GitHub reports conflicts with `dev`. Please sync the latest base and resolve them; the new head will need a fresh CI run and re-review.
   
   **Points from the [previous review](https://github.com/apache/seatunnel/pull/11814#pullrequestreview-5643605276)**
   Please address or reply to each:
   - PR11814-F1 (HIGH): the `received new worker register` wait strategy in `FakeSourceToConsoleWithEventReportIT` is set in `executeExtraCommands`, after the container lifecycle is configured, so it does not reliably gate server readiness. Please move it into the container setup.
   - PR11814-F2 / PR11814-F3 (MEDIUM): when the scheduler fails to terminate, `close()` still flushes the local buffer concurrently with the interrupted-but-running task, `evictAll()` cannot reclaim in-flight connections, and only `HazelcastInstanceNotActiveException`/`IOException` are caught, so other runtime failures escape `close()`. Please serialize the final flush against the scheduled task and broaden the exception handling.
   - PR11814-F4 / PR11814-F6 (MEDIUM/LOW): compile-scope okhttp3 brings a mixed Kotlin stdlib graph (kotlin-stdlib 1.8.21 vs kotlin-stdlib-common 1.9.10) onto the Zeta server classpath, and the same split appears in `known-dependencies.txt` and `seatunnel-dist/release-docs/LICENSE`. Please pin to a single version and confirm the inventories match what the dist actually resolves.
   - PR11814-F7 (LOW): the new `OkHttpClient` follows redirects by default, so configured report headers (often API tokens) could be forwarded to another host. Please disable redirect following or explain why it is acceptable here.
   - PR11814-F8 (LOW): `RequestBody.create(String, MediaType)` appends `; charset=utf-8` when the configured media type has no charset. Please confirm the emitted `Content-Type` stays byte-identical for existing collectors, or build the body from bytes with the exact `MediaType`.
   
   Once the conflicts are resolved and the unit-test job is green, I'll re-review the new head. Thanks!
   
   <!-- streview-comment:2158 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub)" <gi...@apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5754746735

   Thanks for the follow-up commit `416e8cd9d`. A few notes at this head (`416e8cd9d28f10fd976ebaaea5a4b1cea5ac73c4`):
   
   - I haven't been able to verify the earlier points against the diff myself (E2E wait strategy in `FakeSourceToConsoleWithEventReportIT`, `close()` hardening in `JobEventHttpReportHandler`, the Kotlin stdlib split in `known-dependencies.txt` / release-docs `LICENSE`, the unbounded `takeRequest()` waits in `JobEventHttpReportHandlerTest`, redirect following, and the Content-Type charset behaviour), so I'm not marking them resolved yet. Could you point to the commits that address each, or summarise the changes in the PR description? For the Content-Type charset point in particular, please confirm whether the reported header stays byte-identical for existing collectors — I don't see that recorded in the thread.
   
   - CI: the `Build` check for this head and fork run `35491433934` are still `queued`, so there is no CI result for this head yet. At `5cde0fdd8` (fork run `35045349301`), `Dependency licenses`, `Code style` and `engine-v2-it (11, ubuntu-latest)` were green, while `engine-v2-it (8, ubuntu-latest)` stopped in an unrelated module before reaching the console E2E. Once the run for this head finishes, it would be good to have a Java 8 run that actually executes `FakeSourceToConsoleWithEventReportIT` and `HttpReportPackagingIT`.
   
   - There is still an open CHANGES_REQUESTED review on the PR, so I'll hold off on a final call until that is addressed and CI reports for this head.
   
   <!-- streview-comment:1204 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub)" <gi...@apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5770372380

   Thanks @goutamadwant for the quick follow-up. Adding the Public Suffix List attribution to the binary LICENSE/NOTICE and shipping the full MPL 2.0 text is exactly what Issue 3 asked for, and a metadata-only change is the right scope for it.
   
   My approval at `73f66f922` stands; that commit already closed the item from my previous review. Could you point me to the new commit (or confirm it has been pushed to this branch) so I can re-check the release metadata at the new head? I don't see it referenced here yet. Once I've confirmed the attribution and license text in the assembled artifacts, nothing else is outstanding from my side.
   
   <!-- streview-comment:1222 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "goutamadwant (via GitHub)" <gi...@apache.org>.
goutamadwant commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5673461148

   @DanielLeens Update in c13a5357e:
   
   Synced dev and resolved the documentation conflicts while preserving the redirect behavior notes. All seven focused HTTP handler tests pass on Java 11. This also brings in #12302 for the S3 image failure. Fresh CI has been triggered; broader engine E2E results remain pending.


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub)" <gi...@apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5738798275

   Thanks for the follow-up on `5cde0fdd8` — the pom-based reasoning for F4 and the call-order trace for F1 are both helpful.
   
   **F4 (Kotlin/OkHttp pinning).** Agreed that `kotlin-bom:1.9.10` imported into `seatunnel-engine/seatunnel-engine-server/pom.xml` `dependencyManagement` (lines 30-38), together with the versionless `kotlin-stdlib-jdk8` at lines 112-115 and the root-managed `okhttp` (`pom.xml` lines 349-354, `${okhttp.version}`), pins both by construction on this module's classpath. Two remaining asks:
   - Yes, please paste the `mvn dependency:tree -Dincludes=org.jetbrains.kotlin,com.squareup.okhttp3` output for this head so we have the resolved tree on record, not just the pom evidence.
   - Since the runtime graph resolves to 1.9.10, `tools/dependencies/known-dependencies.txt` and `seatunnel-dist/release-docs/LICENSE` should not still carry a 1.8.21 / 1.9.10 split (F6). Please align the inventories with the actual resolved dist and confirm the dependency check passes.
   
   **F1 (wait strategy gating).** The trace makes sense: on the `SeaTunnelContainer.startUp()` → `createSeaTunnelServer(NETWORK)` path, `executeExtraCommands(server)` at line 160 runs before `server.start()` at line 162, so `container.waitingFor(...)` from `FakeSourceToConsoleWithEventReportIT.executeExtraCommands` (lines 106-117) is in effect when `.start()` blocks. I'm fine dropping the HIGH severity on that basis. Your comment was cut off at the `createSeaTunnelContainerWithFakeSourceAndInMemorySink` overload (lines 200-244), where `server.start()` at line 232 precedes `executeExtraCommands` at line 234 — could you finish that thought and confirm `FakeSourceToConsoleWithEventReportIT` never goes through that overload? A one-line comment in the override noting it depends on being applied before `start()` would also help prevent a future regression if someone switches the test to the other factory.
   
   The other items from the earlier review (F2/F3 close() robustness, F5 `takeRequest()` timeout / buffer leak / scheduler-interval assumption in the test, F7 redirect following with report headers, F8 `; charset=utf-8` being appended to the Content-Type) are still open as far as this thread shows — please address them or reply per item so I can track them.
   
   <!-- streview-comment:1164 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub) via commits" <co...@seatunnel.apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-6962806047

   Thanks for the sync, @goutamadwant. Here is where each earlier point stands based on your update.
   
   **Covered by your update**
   - **Redirect restrictions (F7):** noted as already in the branch.
   - **Content-Type (F8):** the note beside the assertion explaining why no charset is appended is what was asked for.
   - **Kotlin stdlib (F4/F6):** noted as aligned at 1.9.10. Could you confirm `known-dependencies.txt` and the release `LICENSE` both show that single version after the merge with `dev`?
   
   **Still need confirmation**
   - **Close path (F2/F3):** do the cancellation/close guards cover these two cases?
     - When the scheduler fails to terminate, does `close()` skip the local-buffer flush rather than run it concurrently with the still-running task?
     - Are runtime failures other than `HazelcastInstanceNotActiveException`/`IOException` kept from escaping `close()`?
   - **E2E wait strategy (F1):** your update doesn't mention it. Is the wait on `received new worker register` now configured before the container starts, rather than in `executeExtraCommands`?
   - **`testRetryAfterHttpFailure` (F5):** also not mentioned. Does it now use a bounded `takeRequest` timeout, release the request buffers, and no longer rely on the scheduler firing immediately despite the 1-day interval?
   
   **Merge gate**
   The earlier approval and Build run 128488973591 were for `06a7b0ac4f416252b4234dd4f8af48d0e9f3b414`. The conflict resolution needs a fresh Build on the head you pushed, plus a check that the OkHttp/Okio relocations and the Hazelcast resource filter are both intact there.
   
   <!-- streview-comment:2419 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "goutamadwant (via GitHub)" <gi...@apache.org>.
goutamadwant commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5576258561

   @DanielLeens Both engine failures occur during worker registration, before the HTTP assertions. Coordinator initialization starts but does not finish. The logs lack a thread dump, so I cannot classify this as unrelated yet. Next is isolated HTTP-enabled/disabled startup with a dump captured before teardown, not another timeout increase.


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "DanielLeens (via GitHub)" <gi...@apache.org>.
DanielLeens commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5551390495

   CI record update: since my last comment (2026-09-02T04:54:40Z), the fork `Build` run I described as "in progress, roughly two-thirds green including the previously-troublesome job" has since completed — and it finished `failure`, not the clean pass I was hoping for.
   
   Specifically, both `engine-v2-it (8, ubuntu-latest)` and `engine-v2-it (11, ubuntu-latest)` failed on the same step: `connector-console-seatunnel-e2e` (the console E2E module, `Failed to execute goal ... maven-failsafe-plugin:verify ... There are test failures`). That module is the one this PR's own diff touches (the `JobEventHttpReportHandler` console E2E test), so I can't currently classify this as an unrelated flake the way I could for the earlier `all-connectors-it-1` failures — it needs an actual look at the failsafe report before this can be called clean.
   
   @goutamadwant could you pull the `connector-console-seatunnel-e2e` failsafe report from that run (`https://github.com/goutamadwant/seatunnel/actions/runs/33579104361`) and check which assertion failed? If it turns out to be a timing/environment flake unrelated to the diff, a rerun should confirm that; if it's a real failure in the console E2E test, that would need to be fixed before this is mergeable. I don't want to leave the earlier "in progress, looking good" framing standing now that the run has actually finished red.


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "goutamadwant (via GitHub)" <gi...@apache.org>.
goutamadwant commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5595786326

   @DanielLeens Added cancellation of outstanding HTTP calls before pool eviction, with a regression that failed before the fix and passes afterward. Aligned the server/starter Kotlin runtime and updated the dependency/license inventories, PR description, and redirect migration notes.
   
   All seven handler tests passed on Java 8 and 11; the 67-module E2E reactor package also passed with test execution skipped.
   
   Isolated Linux startup probes passed with reporting enabled and disabled on both JDKs. They did not reproduce the CI startup timeout, so the E2E now captures thread dumps before teardown without increasing its readiness timeout. New CI is pending.


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub)" <gi...@apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5409350994

   Thanks @DanielLeens for the thorough re-review and for re-tracing the wiring on `a67ad2a93460eabf8f9e672b14fed6a0d926b126` rather than taking the change on faith — much appreciated.
   
   To confirm your reading of the delta: the only change since `8231fd293f24` is the +6/-0 commit to `FakeSourceToConsoleWithEventReportIT.java`, giving the container an explicit 3-minute startup wait in place of the base class's default 60s `Wait.forLogMessage(...)`. This directly follows the diagnostic step you proposed last round for the ambiguous E2E timeout, and it's good to hear the wait strategy is now correctly wired to actually gate readiness — that was the crux of the original finding on this file.
   
   All other items from the earlier rounds (the `close()`/scheduler shutdown hardening, exception handling in the flush path, the okhttp3 4.12.0 migration and its dependency/license registrations, the retry-path regression test, and the redirect/Content-Type behavior questions) remain as fixed in the previous commits, with no changes this round — consistent with your diff stat.
   
   On your one remaining ask: agreed that the change is well-reasoned but not yet confirmed until CI completes on this exact head. I'll make sure a full run goes through on `a67ad2a93460eabf8f9e672b14fed6a0d926b126` and post back here once it finishes, with particular attention to whether the event-report E2E test passes reliably under the extended wait. If it does, I'd ask you to take one final look to close out the review; if the timeout still reproduces, we'll know the wait duration wasn't the root cause and can dig further.
   
   Thanks again for the careful, multi-round review — it made this migration much safer.
   
   <!-- streview-comment:561 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub)" <gi...@apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5747123570

   Thanks @goutamadwant for the update — good to hear all seven handler tests pass on Java 11 locally.
   
   **F4 / F6 (Kotlin & OkHttp pinning).** The pom trace earlier in this thread (`kotlin-bom:1.9.10` imported into the engine-server module's `dependencyManagement`, `okhttp` managed at the root via `${okhttp.version}`) addresses the pinning concern. Since the resolved tree hasn't actually been posted here yet, could you paste the `mvn dependency:tree -Dincludes=org.jetbrains.kotlin,com.squareup.okhttp3` output against the current head? If it confirms a uniform 1.9.10 graph, I'll consider F4 resolved, and please align the license/dependency inventory entries that still list 1.8.21 so they match the actual dist.
   
   **F1 (E2E wait strategy).** The call-order trace for the `startUp()` → `createSeaTunnelServer(NETWORK)` path, where `executeExtraCommands(server)` runs before `server.start()`, is convincing. Since `createSeaTunnelContainerWithFakeSourceAndInMemorySink` has the opposite order, could you either confirm `FakeSourceToConsoleWithEventReportIT` only ever goes through the `startUp()` path, or add a short comment in the override noting that assumption? With that I'm fine closing F1.
   
   **Still open** (since the handler patch is unchanged, these haven't moved):
   - **F2 / F3** – in the handler's close(): when the scheduler doesn't terminate, the local buffer is still flushed concurrently with the interrupted task, and the flush only catches `HazelcastInstanceNotActiveException`/`IOException`. Please guard the flush on successful termination (or drain under the same lock) and widen the catch so other runtime failures don't escape close().
   - **F5** – `testRetryAfterHttpFailure`: use the timeout variant of takeRequest so the build can't hang, close the request Buffers, and make the scheduler trigger explicit rather than relying on it firing despite the 1-day interval.
   - **F7** – please disable redirect following on the new OkHttpClient so configured report headers/tokens aren't forwarded to a redirect target on another host.
   - **F8** – confirm whether `RequestBody.create(String, MediaType)` appending `; charset=utf-8` to the Content-Type is acceptable for existing collectors, or build the body from bytes to keep the header byte-identical.
   
   I'll take another look once the queued CI run finishes.
   
   <!-- streview-comment:1179 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "DanielLeens (via GitHub)" <gi...@apache.org>.
DanielLeens commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-6266725003

   Thanks @SEZ9 for going back to the actual head. No new commit has landed since my last review at `131de7f704`, so I re-read the live source at that exact SHA and am pointing at the lines for each item (all in the PR diff, `JobEventHttpReportHandler.java` unless noted):
   
   1. **F1 - E2E wait.** `FakeSourceToConsoleWithEventReportIT.java:113-125` sets `container.waitingFor(...)` with the `.*received new worker register:.*` regex from inside `executeExtraCommands` (starts at line 95). The comment at lines 111-112 records the ordering: this test goes `startUp() -> createSeaTunnelServer()`, which calls the hook *before* `start()`, so the strategy does gate container startup. The wrapper only adds a thread dump (`logStartupThreads`, line 129) on a failed wait and rethrows.
   2. **F2/F3 - `close()`.** Lines 214-270. `closing` is set first (line 216) and `handle()` returns early on it (line 125). The final flush at line 238 runs only if `schedulerTerminated && !interrupted`; otherwise it is skipped with a warning (line 261), so the buffer flush cannot overlap a still-running task. `finally` (lines 263-269) always runs `dispatcher().cancelAll()` and `connectionPool().evictAll()` and restores the interrupt flag. Wrapped failures are unwrapped via `CompletionException` (lines 243-246); only `HazelcastInstanceNotActiveException` is logged at INFO, everything else goes to ERROR (lines 247-253), and the local buffer flush runs regardless (line 256). Tests: `testInterruptedCloseCancelsInFlightRequest` (test:127), `testCloseWhenHazelcastNotActive` (test:347), `testCloseLogsUnexpectedRingbufferFailure` (test:398).
   3. **F4/F6 - Kotlin.** `seatunnel-engine-server/pom.xml` imports `kotlin-bom` (line 34) and declares `kotlin-stdlib-jdk8` without a version (line 114), so the whole stdlib family resolves to one version. `known-dependencies.txt` lines 77-80 list `kotlin-stdlib`, `-common`, `-jdk7`, `-jdk8` all at `1.9.10`, and `release-docs/LICENSE` lines 258-261 list the same four at `1.9.10` (okhttp/okio at 262-264).
   4. **F5 - retry test.** `testRetryAfterHttpFailure` (test:207) uses `Duration.ofDays(1)` (line 218) so no periodic tick can drive the retry, bounds both `takeRequest` calls at 10 seconds (lines 223-224), and both recorded request `Buffer`s are closed via try-with-resources (lines 227-230).
   5. **F7 - redirects.** Lines 284-285: `.followRedirects(false)` and `.followSslRedirects(false)`; `testDoesNotFollowRedirects` (test:238) asserts the redirect target never receives a request.
   6. **F8 - Content-Type.** `postEvents` (lines 192-196) uses `RequestBody.create(events, httpMediaType)` with `MediaType.parse("application/json")`. `testReportEvent` asserts the header is `application/json; charset=utf-8` (test:190); okhttp3 adds the charset for a String body the same way the 2.x call did, so existing collectors see the same header. The user-visible behaviour changes that remain (redirects not followed, modern TLS defaults only) are documented in `incompatible-changes.md` (en and zh) and `event-listener.md`.
   
   On the CI side, the required `Build` is `SUCCESS` for this head. I did not run any of this locally; the above is a source-level read at `131de7f704`. @goutamadwant nothing further is needed from you on these items from my side.
   
   Please let me know if any line reference does not match what you see on your side.


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "goutamadwant (via GitHub) via commits" <co...@seatunnel.apache.org>.
goutamadwant commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-6712370598

   Resolved the dev conflicts and verified that the runtime Kotlin libraries still resolve to 1.9.10 through the BOM.


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub) via commits" <co...@seatunnel.apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-6817018444

   @goutamadwant thanks for the point-by-point walkthrough, and sorry for the delayed reply.
   
   On F1–F6, your explanations line up with what I was after:
   
   - **F1** – If `executeExtraCommands(server)` runs before `server.start()` in `createSeaTunnelServer`, the wait strategy set in the hook does gate readiness, and the config copy in the same hook relying on that order is a good confirmation.
   - **F2** – Running the final flush only when the scheduler terminated and the thread wasn't interrupted, plus cancelling in-flight calls before `evictAll()` in `finally`, is what I was looking for.
   - **F3** – Catching `Exception` on both flush paths and restoring the interrupt flag, with test coverage in `JobEventHttpReportHandlerTest`, addresses the concern.
   - **F4** – Importing `kotlin-bom` 1.9.10 in `seatunnel-engine-server/pom.xml` resolves the mixed stdlib graph, and your 11-07 note that the runtime Kotlin libraries still resolve to 1.9.10 through the BOM after the `dev` merge is good to have on record.
   - **F5** – Explicit `report()` call, 10s bound on `takeRequest`, and closing the buffers covers all three parts.
   - **F6** – Thanks for aligning the Kubernetes block in `release-docs/LICENSE` to 1.9.10 in `fbd73878d6`; with `known-dependencies.txt` already at 1.9.10 only, the inventories should now match.
   
   I'll confirm each of these against the diff on the current head `06a7b0ac` and resolve them once verified.
   
   Your comment looks like it was cut off after F6, so two items are still open:
   
   - **F7** – Does the new `OkHttpClient` still follow redirects by default? If so, please either disable that or explain why forwarding the configured report headers to a redirect target on another host is acceptable here.
   - **F8** – Can you confirm whether the `Content-Type` header produced by `RequestBody.create(String, MediaType)` stays byte-identical to the previous implementation when the configured media type has no charset? If it changes, a short note in the PR description for existing collector users would be enough.
   
   Once F7 and F8 have a reply I'm happy to approve.
   
   <!-- streview-comment:2282 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub) via commits" <co...@seatunnel.apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-6889159189

   Thanks for the follow-up. A quick status on the open points:
   
   **Merge state.** Head `06a7b0ac4f416252b4234dd4f8af48d0e9f3b414` is reported as conflicting with `dev`, even though the [Build](https://github.com/apache/seatunnel/runs/128488973591) is green for that commit. Could you resolve the conflict against the latest `dev` and push, so the checks run on the resolved head?
   
   **F7 / F8.** For F8, could you point me at the assertion (or diff) that pins the `Content-Type` wire behavior when the configured media type has no charset? A short comment next to it stating the intent would also help keep it from being removed later. For F7, could you confirm whether the `OkHttpClient` is built with redirects disabled, or explain why forwarding the configured report headers on redirect is acceptable?
   
   **F1–F6.** I don't see changes for these on the current head, so a short per-item status would help:
   - F1: making the `received new worker register` wait strategy actually gate container startup instead of running in `executeExtraCommands`.
   - F2 / F3: `close()` behavior when the scheduler doesn't terminate — guarding the local-buffer flush against the still-running task, and handling failures other than `HazelcastInstanceNotActiveException`/`IOException` so they don't escape `close()`.
   - F4 / F6: aligning the Kotlin stdlib versions (1.8.21 vs 1.9.10) in the dependency and license inventories, and confirming they match what the dist actually resolves.
   - F5: a bounded `takeRequest()` in `testRetryAfterHttpFailure`, closing the request `Buffer`s, and triggering the scheduler explicitly rather than relying on it firing under a 1-day interval.
   
   If any of these are already handled, just point me at the change and I'll verify on the new head. Otherwise, happy to review them in a follow-up commit on this PR.
   
   <!-- streview-comment:2367 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "DanielLeens (via GitHub)" <gi...@apache.org>.
DanielLeens commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5412549401

   Thanks for the recap, @SEZ9 — your read of the delta since my last review is accurate: the only change on `a67ad2a93460eabf8f9e672b14fed6a0d926b126` since my round-4 comment is the `FakeSourceToConsoleWithEventReportIT` startup-wait extension, and everything else (license/dependency bookkeeping, the `close()` pool-eviction `finally`, the retry-path test) is unchanged from what I already re-traced and confirmed in earlier rounds.
   
   One correction on the CI front, since that's the one open item: I checked the fork run for this exact head directly rather than relying on the "Build" label — `goutamadwant/seatunnel` run `32676810603` (the one covering `a67ad2a93460eabf8f9e672b14fed6a0d926b126`) shows `conclusion: cancelled`, not a completed pass or fail. Every job in it, including `engine-v2-it` (the one that runs `FakeSourceToConsoleWithEventReportIT`, the test this round's fix targets), is `cancelled` rather than `success`/`failure`. So we still don't have an actual completed run to confirm the extended wait fixed the timeout — the run that would tell us that got cancelled before finishing, not "finished" as I'd read your message to suggest.
   
   @goutamadwant — could you push an empty commit or otherwise trigger a fresh run on this head so we get one that actually completes? Once `engine-v2-it` finishes (pass or fail) on this exact SHA, I'll take a final look. Nothing in the source review is blocking at this point — this is purely about getting a completed CI signal before 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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "DanielLeens (via GitHub)" <gi...@apache.org>.
DanielLeens commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5379439601

   Thanks for chasing this down carefully instead of just retrying blindly - that's the right instinct given it's the one test file this PR touches.
   
   I pulled the raw logs for all four fork attempts on this head (run 31960889724, attempts 1-4) myself to check your read. Confirming what you found: in every attempt, `FakeSourceToConsoleWithEventReportIT` fails at `SeaTunnelContainer.createSeaTunnelServer` with `ContainerLaunchException: Timed out waiting for log output matching '.*received new worker register:.*'` after ~63-64s, and the captured container log stream has **zero lines** in that window - not even a JVM startup banner, let alone a Hazelcast/worker-register line.
   
   Two things stood out when I compared this against the rest of the same job:
   
   1. Elsewhere in the exact same `engine-v2-it` job, the identical `seatunnelhub/openjdk:8u342` image starts and logs "received new worker register" successfully **147 times**, each taking about 6-8 seconds. So the image itself, the Docker pull, and the base container-boot path are all clearly working fine on this runner during this run - it's isolated to this one test.
   
   2. I read `createHttpClient()` in your current `JobEventHttpReportHandler` (lines ~251-263): it's a pure in-memory `OkHttpClient.Builder().build()` call - no DNS, no proxy probing, no socket I/O at construction time. So I don't think your new code is *itself* capable of blocking the container's JVM for 60+ seconds. If the extra okhttp3/kotlin-stdlib classes on the classpath were slowing things down, I'd expect to still see early JVM/log4j startup output before any hang, not total silence for the entire window.
   
   That combination (zero output, not partial-then-stall) reads to me as more consistent with the container/runner not getting scheduled promptly at the Docker-daemon level (resource contention on the runner, similar in spirit to the Windows Hazelcast "Node failed to start!" flake and the Couchbase container-bootstrap timeouts we've seen intermittently on unrelated PRs) than with a code-level regression in this PR's diff - but I want to be upfront that I can't fully rule your diff out from static/log analysis alone, since I don't have a `dev`-branch control run of this exact test on this exact runner to compare against.
   
   Concrete next step I'd suggest, cheapest first: temporarily bump the wait-strategy timeout for just this one IT (or attach a log consumer from container start rather than only from the wait strategy) and do one more fork run. If the container eventually does emit a startup banner and register message, just later than 63s, that confirms plain scheduling/resource contention and a rerun or a slightly longer timeout is all that's needed - no diff change required. If it stays completely silent even with a much longer window, that would point at something more structural (networking/image) and would be worth flagging separately from this PR's dependency change, since your code path doesn't do anything at construction time that should cause total silence.
   
   This is not a known, previously-documented flake signature for this specific test on my side, so I can't just tell you "yes, that's a known issue, ignore it" - but based on the above I also don't think it's your fix that's causing it. I'd treat one more attempt with better diagnostics as the fastest way to get a real answer rather than guessing further from logs alone.
   


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "goutamadwant (via GitHub)" <gi...@apache.org>.
goutamadwant commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5305615346

   @DanielLeens Completed the okhttp3 migration follow-up. Added the runtime dependency and license entries, release the private connection pool on close, and added a regression that verifies the same payload is retried after a 500 response. The focused suite passes with 3 tests.


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "DanielLeens (via GitHub)" <gi...@apache.org>.
DanielLeens commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5662012668

   @SEZ9 — I think this crossed with my comment from about 11 minutes earlier on this same thread, where I went through F1-F8 individually against this exact head (`d9855aee7d7`) and found all eight already resolved, each with a specific line pointer. Restating them here directly so there's no ambiguity about what's already on the branch:
   
   - **F1 (wait-strategy ordering)** — `SeaTunnelContainer.java`: `createSeaTunnelServer(Network)` calls `executeExtraCommands(server)` at line 147 and only calls `server.start()` afterward at line 149. `FakeSourceToConsoleWithEventReportIT.startUp()` resolves through that exact overload, and its `executeExtraCommands` override installs the `LogMessageWaitStrategy` there — before `start()` runs, so it does reliably gate readiness on this test's actual call path.
   - **F2/F3 (`close()` hardening)** — `JobEventHttpReportHandler.java:213-259`: both final flushes are gated on `schedulerTerminated && !interrupted` (`:237`) with a logged `else` branch instead of racing an interrupted/still-running scheduler; both flushes are wrapped in `catch (Exception e)` (`:241`, `:246`), not the old narrow `HazelcastInstanceNotActiveException`/`IOException` pair; `dispatcher().cancelAll()` runs before `connectionPool().evictAll()` in the outer `finally` (`:253-254`), so in-flight calls are cancelled before their connections are evicted.
   - **F4/F6 (Kotlin stdlib split)** — `known-dependencies.txt` lists `kotlin-stdlib{,-common,-jdk7,-jdk8}` all at `1.9.10` (no `1.8.21` remaining); `seatunnel-engine-server/pom.xml:30-37` imports `kotlin-bom:1.9.10` in `dependencyManagement`, pinning the whole graph instead of leaving it to Maven nearest-wins.
   - **F5 (`testRetryAfterHttpFailure` hygiene)** — `JobEventHttpReportHandlerTest.java:208-209`: both `takeRequest` calls use an explicit `10, TimeUnit.SECONDS` bound; the redirect test at `:247`/`:252` follows the same pattern.
   - **F7 (redirect-following)** — `JobEventHttpReportHandler.java:264-275`: `createHttpClient()` builds with `.followRedirects(false).followSslRedirects(false)`, so a 3xx surfaces through the existing non-2xx/retry/log path instead of forwarding configured report headers to a redirect target on another host.
   - **F8 (Content-Type charset)** — `JobEventHttpReportHandlerTest.java:175`: `testReportEvent` asserts the emitted header is exactly `"application/json; charset=utf-8"`, locking the migration's behavior in by assertion rather than just review claim.
   
   I also independently verified the merge itself changed none of this: `git diff af713e463f3..d9855aee7d7 -- <every file this PR's diff touches>` is empty for the handler, its test, and all three POMs — the only PR-owned files the merge commit touches are the `incompatible-changes.md` docs (en+zh), where the resolution is pure content-preservation (three unrelated `dev` entries appended alongside this PR's own, nothing removed or altered). So this really is a mechanical sync, and F1-F8 were already fixed on `af713e463f3` before it landed — same conclusion you reached independently in your comment just above.
   
   If you're seeing something different on your end for any of these (e.g. a different line range or a case I'm not tracing correctly), point me at the specific spot and I'll re-check it directly — but from current source I don't have anything left open here.
   
   On CI: as of this comment the fork's `Build` run for `d9855aee7d7` is still queued/in-progress, so I don't have a completed result on this exact merge commit yet. Given the merge is a no-op on every file this PR touches, I'd expect it to match the previous head's result (both `engine-v2-it` legs green, with the `minio/minio` registry failures elsewhere being an unrelated, independently-observed outage that night) — but that's an expectation, not a confirmed result, so please don't treat this as a green signal until the run actually completes.
   


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "goutamadwant (via GitHub)" <gi...@apache.org>.
goutamadwant commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5808313985

   For F5/F7/F8 at the current head ec6a7f9affc: `testRetryAfterHttpFailure` calls `handler.report()` twice, bounds both `takeRequest` calls at 10 seconds, and closes both request buffers. The one-day interval suppresses another periodic tick; the test does not wait for it to drive the retry. Redirect following is disabled with both `followRedirects(false)` and `followSslRedirects(false)`, and `testDoesNotFollowRedirects` verifies the target receives no request. One correction to the prior thread on F8: the current `testReportEvent` asserts `Content-Type: application/json; charset=utf-8`. The pre-PR OkHttp 2.7.5 `RequestBody.create(MediaType, String)` also appends `; charset=utf-8` when the media type lacks a charset, so that header is preserved, not a new incompatibility. The 11 focused reporter tests pass.


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "DanielLeens (via GitHub)" <gi...@apache.org>.
DanielLeens commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5812745144

   Thanks both — and @goutamadwant, thanks for catching this. I want to correct something I got wrong in my own comment from earlier today rather than let it stand.
   
   **F8 correction.** I previously said the emitted `Content-Type` was byte-identical with no `; charset=utf-8` suffix, reasoning only from `MediaType.parse("application/json")` carrying no charset parameter. That reasoning was incomplete — I looked at the `MediaType` alone and didn't trace what the `RequestBody` factory does with it. Re-checking the actual call at `JobEventHttpReportHandler.java:196` (`RequestBody.create(events, httpMediaType)`, the `String` + `MediaType` overload): when the supplied `MediaType` has no charset, okhttp3's `RequestBody.create(String, MediaType)` sets one to UTF-8 and rewrites the content type to include `; charset=utf-8` before it's ever attached to the request. That's exactly what `JobEventHttpReportHandlerTest.java:190` asserts (`"application/json; charset=utf-8"`), so the test is correct and my prior claim was not.
   
   I also checked the pre-PR behavior directly rather than taking @goutamadwant's word for it: at the merge-base commit (before this PR), `JobEventHttpReportHandler` used `com.squareup.okhttp:mockwebserver:2.7.5` (see the old `pom.xml`) and called `RequestBody.create(httpMediaType, events)` — the equivalent okhttp 2.x overload, which has the identical well-known behavior of appending `; charset=utf-8` when the passed `MediaType` doesn't specify one. So this is preserved legacy behavior, not a new incompatibility introduced by the migration — @goutamadwant's answer is correct on both counts (current behavior and why it matches the legacy client).
   
   **F5 and F7, independently re-verified against `ec6a7f9affc` (no new commit, so this is the same source I reviewed before):**
   - F5: `testRetryAfterHttpFailure` (`JobEventHttpReportHandlerTest.java`) calls `handler.report()` twice explicitly rather than waiting on the `Duration.ofDays(1)` interval to fire, bounds both `takeRequest` calls at `10, TimeUnit.SECONDS`, and closes both request `Buffer`s via try-with-resources. Matches the description.
   - F7: `createHttpClient()` still builds with `.followRedirects(false).followSslRedirects(false)`, and `testDoesNotFollowRedirects` exercises it with a real 307 response. Unchanged from what I confirmed in earlier rounds.
   
   With F8 now correctly resolved (preserved, not a new incompatibility) and F5/F7 re-confirmed directly against source, I don't have anything further open from my side. No new commit has landed since `ec6a7f9affc`, which is the head my `APPROVED` review already covers, so that approval stands — this comment is a factual correction plus re-verification, not a new review.
   


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "DanielLeens (via GitHub)" <gi...@apache.org>.
DanielLeens commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5713116411

   Thanks for the careful comparison, @SEZ9 — appreciate you re-verifying against the sync commit rather than taking it on faith.
   
   **1. CI on the new head (5cde0fdd8).** I pulled this from `goutamadwant/seatunnel` run `35045349301` in my last review, and just re-confirmed both jobs directly via the Actions API for this exact head:
   - `engine-v2-it (11, ubuntu-latest)`: `success`. This is the leg you asked about — it gets through the engine module and into `connector-console-seatunnel-e2e`, running both `FakeSourceToConsoleWithEventReportIT` and `HttpReportPackagingIT`, both passing (raw surefire output quoted in my last review).
   - `engine-v2-it (8, ubuntu-latest)`: `failure`. I re-pulled the full job log for this run (job `104636467785`) just now and traced it directly: the reactor plan lists `SeaTunnel : E2E : Engine : Console` as module `[67/67]`, but the log never reaches a "Building SeaTunnel : E2E : Engine : Console" line — module `66/67` (`connector-seatunnel-e2e-base`) fails on `SplitClusterFaultToleranceIT.testStreamJobCancelResolvesWhenWorkerCrashesBeforeCancelAck` (`ConditionTimeout`, unrelated to HTTP event reporting), and the Maven reactor (no `-fae`) stops before module 67 ever starts. So on JDK 8, `connector-console-seatunnel-e2e` is genuinely skipped, not failed — same picture as before, now re-verified against this exact head's log rather than carried forward.
   
   **2. Kotlin stdlib versions (F4/F6).** I just re-pulled both files directly from the GitHub Contents API at this exact head (`5cde0fdd8ce8ace10b688010f968bfeeee61d6b0`), not from memory:
   - `tools/dependencies/known-dependencies.txt`: `kotlin-stdlib-1.9.10.jar`, `kotlin-stdlib-common-1.9.10.jar`, `kotlin-stdlib-jdk7-1.9.10.jar`, `kotlin-stdlib-jdk8-1.9.10.jar` — four entries, all `1.9.10`.
   - `seatunnel-dist/release-docs/LICENSE`: the same four `kotlin-stdlib*` entries, all `1.9.10`.
   
   I don't see a `1.8.21` entry in either file on this head. Could you point me to the specific file/line where you're seeing `kotlin-stdlib` at `1.8.21`? It's possible you're reading a cached or earlier view — `af713e463f3`/`d9855aee7d7` already had this pinned uniformly to `1.9.10` when I checked them in prior rounds too — but I'd like to check the exact source rather than assume.
   
   Given both: for (1), the JDK 8 gap is a pre-existing, already-tracked flake (open PR #12311, open coverage PR #12030) unrelated to this PR's diff — I wouldn't hold the merge open indefinitely for it to happen to line up on JDK 8 specifically, since JDK 11 already gives a real, completed positive signal for the exact tests this PR owns. For (2), the inventories look correctly aligned as-is on the current head — happy to revisit if you can point me to where you're seeing the split.


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "goutamadwant (via GitHub)" <gi...@apache.org>.
goutamadwant commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5743951916

   Merged dev's NebulaGraph/RocketMQ fixes; all seven HTTP handler tests pass on Java 11. The engine-server dependency tree confirms OkHttp 4.12.0 and Kotlin stdlib variants 1.9.10. The HTTP migration patch is unchanged, and fresh CI is queued.


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "goutamadwant (via GitHub)" <gi...@apache.org>.
goutamadwant commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5770123134

   @DanielLeens Added the missing Public Suffix List attribution to the binary LICENSE/NOTICE and included the full MPL 2.0 license. Verified both binary assembly descriptors include these files; Java 11 parent verification passed. This follow-up changes release metadata only. 


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "SEZ9 (via GitHub)" <gi...@apache.org>.
SEZ9 commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5852114232

   Thanks for the detailed re-check against `ec6a7f9affc`.
   
   **F1, F2/F3, F4/F6.** Your description of the call order in `executeExtraCommands`, the `close()` structure, and the `kotlin-bom` import is helpful and consistent with what would address these points. Since no new commit has landed since `ec6a7f9affc`, I'll re-verify these directly against the changed files at that head rather than closing them on the description alone, and I'll update the findings once that's done. Your comment also appears to be cut off after "no `1.8`" — no need to repost, I'll check the inventory files myself.
   
   **Still open — not covered in this update:**
   - **F5** (retry test): is `takeRequest()` now called with a timeout, are the request buffers closed, and how does the test trigger the retry rather than relying on the 1-day scheduler interval?
   - **F7** (redirects): has redirect following been disabled on the `OkHttpClient`, or is there a reason forwarding the configured report headers on redirect is acceptable here?
   - **F8** (`Content-Type`): is the emitted header unchanged for existing collectors, or is the `; charset=utf-8` suffix intentional?
   
   A short pointer for each of F5/F7/F8 (or a note that they'll land in a follow-up) is all that's needed to wrap this up.
   
   <!-- streview-comment:1355 -->


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "DanielLeens (via GitHub)" <gi...@apache.org>.
DanielLeens commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-5577170135

   Thanks for digging into this, @goutamadwant — that's a useful data point. If both `engine-v2-it (8)` and `engine-v2-it (11)` are stalling during worker registration/coordinator init and never getting as far as the `JobEventHttpReportHandler` HTTP assertions, that does point away from a defect in this PR's own `JobEventHttpReportHandlerTest` logic and toward either (a) a genuine startup-timing regression somewhere in the bootstrap path this PR's `okhttp3` migration touches indirectly (e.g. classloading/dependency shading order affecting cluster join), or (b) an environment/resource-contention issue on the runner that's unrelated to the diff — but I agree with you that we can't respectably call it either one without a thread dump. Guessing at this point (e.g. bumping a timeout again) would just mask the real failure mode if it recurs after merge.
   
   Your next step — an isolated HTTP-enabled vs. HTTP-disabled startup run with a dump captured before teardown — is exactly the right way to isolate whether this is specific to the `okhttp3` client initialization holding up worker registration, or a pre-existing flake in the coordinator bootstrap that would reproduce with or without your change. Please share the dump (or a link to the run) once you have it either way.
   
   No new commit has landed since my last full review (head is still `dfda45aef`), so I'm not re-reviewing the code again in this comment — my prior conclusions on `JobEventHttpReportHandler.java` and its test stand as posted. I'll hold off on any merge-readiness call until we know whether this `engine-v2-it` failure is diff-related; once you've got the isolated dump, or once a fix/adjustment actually lands, ping me and I'll take it from there — a full re-review if there's a new commit, or a direct answer on the CI question if it's just new evidence.


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "DanielLeens (via GitHub) via commits" <co...@seatunnel.apache.org>.
DanielLeens commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-6877328282

   Thanks, @SEZ9, for checking the F7/F8 evidence. The charset assertion already pins the pre-existing wire behavior; a short note beside it would make that intent clear. I’ll keep your F1–F6 follow-up separate from my technical approval. GitHub still reports a content conflict against dev, so the conflict resolution will need a fresh Build and a check that those paths remain intact on the resulting head.


-- 
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] [Improve][Engine] Migrate job event HTTP reporting to okhttp3 [seatunnel]

Posted by "goutamadwant (via GitHub) via commits" <co...@seatunnel.apache.org>.
goutamadwant commented on PR #11814:
URL: https://github.com/apache/seatunnel/pull/11814#issuecomment-6651167489

   @davidzollo thanks for the approval. I have answered SEZ9's remaining items above with the code pointers; the only new commit updates three LICENSE lines. Could you merge when convenient?


-- 
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