Caideyipi opened a new pull request, #19193:
URL: https://github.com/apache/iotdb/pull/19193

   ## Description
   
   Backport the applicable changes from #19160 
(c5f57e9305e795fbd97d6810f3234aed91a2cf39) onto `dev/1.3` at 
8f589d4b153d7cf648a8228475e27a2085db6ed9.
   
   Non-query test helpers now close statements and internally created 
connections on success and failure, while preserving caller-owned connections. 
The Pipe diagnostics tests wait for the asynchronous leader-change procedure 
and its coordinator lock before issuing DROP, and verify that DROP creates a 
new connection attempt.
   
   Compatibility adaptations:
   
   - Keep the existing `dev/1.3` retry helper, which already uses 
try-with-resources. Adapt its cleanup regression scenarios to the existing 
four-argument API; there is no borrowed-connection retry overload on this 
branch.
   - Match the existing `dev/1.3` leader-change log marker and retain the 
tree-model test categories.
   - Intentionally omit the rejected-submission stubbing hunk: this branch 
lacks the corresponding regression test and prerequisite rollback behavior from 
#19119. No prerequisite product changes are included.
   
   ## Validation
   
   - Spotless passed with the integration-test profile enabled.
   - Built the matching TsFile dependency from source. A clean ClusterIT 
reactor executed all 711 goals without output-cache reuse; the required IoTDB 
modules and integration-test sources compiled successfully.
   - All three new resource-cleanup tests passed under both SimpleIT and 
ClusterIT (no failures, errors, or skips). The SimpleIT dependency build partly 
reused cached outputs; the subsequent ClusterIT build rebuilt the required 
dependency closure from clean source.
   - The clean ClusterIT attempt passed the Pipe recovery scenario but failed 
the existing concurrent-operation assertion: ALTER returned 
`PIPE_NOT_EXIST_ERROR` (1806) instead of `PIPE_ERROR` (1107). The server log 
shows that the background DROP finished immediately after its timeout response, 
before ALTER ran. The new leader-change barrier and connection-count assertion 
passed. No assertions or product behavior were changed to hide this result.
   - A single focused retry of the concurrent-operation scenario reproduced the 
same 1806-versus-1107 failure. The DROP completion flow and timeout 
configuration are also present in the source revision; this remaining timing 
race needs a separate follow-up.
   


-- 
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