CRZbulabula commented on PR #17934:
URL: https://github.com/apache/iotdb/pull/17934#issuecomment-4704352086
Fixed the CI failure in \`IoTDBRemoveDataNodeRegionAllocationIT\` (e75e196).
**Root cause**: after \`kill -9\`-ing the target DataNode, the test
re-opened a multi-node \`EnvFactory.getConnection()\`. That connection fans
every read out to *all* DataNodes, including the dead one, so each
\`executeQuery\` (\`SHOW DATANODES\`, region maps) returned \`[Connection
Error, null, null, null]\` and failed with \`InconsistentDataException\` at the
connection site — before the test could exercise the allocation behaviour.
**Fix**: pin all post-kill SQL (both the main statement and the
allocation-probe connection) to a surviving DataNode via a new
\`BaseEnv.getConnection(DataNodeWrapper)\` overload, which scopes write *and*
read to a single live node. This matches the existing framework intent ("This
is useful when you shut down a dataNode").
The remaining CI reds (\`IoTDBConnectionsIT\` on Windows,
\`IoTDBPipeClusterIT\` on dual-table-manual-enhanced) are unrelated
pre-existing flakes — they touch the session and pipe subsystems, not region
allocation.
--
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]
