Caideyipi commented on PR #19037:
URL: https://github.com/apache/iotdb/pull/19037#issuecomment-6487439232

   Independent re-review of the full PR diff at 
`8df231b051cdacb0c7f2163591077a9f276d0c29`, including one-way template 
matching, shared-cluster isolation, internal consensus Pipe preservation, 
asynchronous Pipe deletion waits, and receiver cleanup. This single round found 
two P2 test-isolation issues, fixed in 
`904b94a1fac80fda6bb293ceef0d99e5f03d0754` and 
`ac51078cf24c43757493520a22af4bb4b926932d`.
   
   - **Close both subscription consumers on exceptional paths.** In 
[`IoTDBSubscriptionIsolationIT.java:165`](https://github.com/apache/iotdb/blob/8df231b051cdacb0c7f2163591077a9f276d0c29/integration-test/src/test/java/org/apache/iotdb/subscription/it/local/tablemodel/IoTDBSubscriptionIsolationIT.java#L165),
 an RPC/assertion failure could bypass the trailing consumer closes. Active 
subscriptions then prevent shared-cluster teardown from dropping topics and 
leave state for subsequent methods. Both consumers now use try-with-resources; 
the assertions and explicit unsubscribe checks remain.
   - **Preserve protected system audit databases during cleanup.** 
[`TestUtils.clearAllDatabases`](https://github.com/apache/iotdb/blob/8df231b051cdacb0c7f2163591077a9f276d0c29/integration-test/src/test/java/org/apache/iotdb/db/it/utils/TestUtils.java#L1106)
 and the duplicate cleanup in 
[`AbstractSubscriptionConsensusLocalIT.java:115`](https://github.com/apache/iotdb/blob/8df231b051cdacb0c7f2163591077a9f276d0c29/integration-test/src/test/java/org/apache/iotdb/subscription/it/consensus/local/AbstractSubscriptionConsensusLocalIT.java#L115)
 attempted to delete protected audit databases. The subscription-enabled run 
reproduced SQL 803 teardown errors; Apache sources define the same audit 
databases and deletion restrictions. The shared helper now preserves the 
appropriate tree/table audit database using existing constants, and the 
consensus fixture reuses shared subscription cleanup.
   
   Actual verification in this round:
   
   - `mvn spotless:apply -P with-integration-tests` and `git diff --check` 
passed after the fixes.
   - At exact `8df231b051c`, `IoTDBShowReceiversIT` under `MultiClusterIT1`: 
**9 tests, 0 failures/errors/skips**. Matching TsFile and all required Apache 
IoTDB reactor modules were rebuilt from clean source with compiler caches 
disabled. This closes the previous extension-resolution blocker; it is an 8df 
result, not a receiver rerun at ac51078.
   - At `ac51078cf24`, using a corresponding subscription-enabled source build 
with the current PR test fixtures/children and cleanup helpers overlaid: 
**SimpleIT: 10 tests, 0 failures, 0 errors, 0 skips**; **TableClusterIT: 2 
tests, 0 failures, 0 errors, 0 skips**. Both Maven verification commands exited 
0.
   - The enabled-edition build retains `setSubscriptionEnabled(true)` in the 
fixtures. The committed PR cleanup-helper block is used exactly, with no 
verification-only audit filtering. Required IoTDB modules were clean-rebuilt 
for each profile with compiler caches disabled; the unchanged matching TsFile 
source build was reused with recorded provenance. These are corresponding 
enabled-edition runs, not OSS subscription results 
(`SUBSCRIPTION_ENABLED=false` and the OSS exclusion remain).
   - Before the audit fix, the exact PR SimpleIT selection had **10 tests, 6 
teardown errors, 0 assertion failures/skips**; these are failed results. An 
earlier enabled-edition attempt had 1 error and 4 skips. Neither is counted as 
a passing subscription suite.
   - The added path/template regression methods were not rerun in this round 
and have not run locally. Their earlier remote record at `de2c7d730951` remains 
historical: `PartialPathTest` 19 tests and `ClusterTemplateManagerTest` 4 tests.
   
   The remote head is `ac51078cf24c43757493520a22af4bb4b926932d`, and the PR 
remains **draft**. This round found test-logic issues, so successful 
fixes/checks do not constitute an issue-free re-review. A fresh session must 
review the updated full diff; no second review round was performed here.
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to