hudi-agent commented on code in PR #19812:
URL: https://github.com/apache/hudi/pull/19812#discussion_r3937709418
##########
hudi-sync/hudi-hive-sync/src/main/java/org/apache/hudi/hive/ddl/HiveQueryDDLExecutor.java:
##########
@@ -77,27 +77,32 @@ public HiveQueryDDLExecutor(HiveSyncConfig config,
IMetaStoreClient metaStoreCli
this.metaStoreClient = metaStoreClient;
this.driverPool = driverPool;
this.metaStoreClientPool = metaStoreClientPool;
+ // SessionState.start() attaches the session it starts to this thread,
displacing whatever the
+ // caller had there -- another executor's session, or that of an
application embedding this
+ // sync. Ours is not the thread's to keep: every statement and the
teardown bind it
+ // explicitly, so give the thread back once the Driver, whose constructor
reads
+ // SessionState.get(), has been built.
+ SessionState previousSession = SessionState.get();
+ ClassLoader previousLoader =
Thread.currentThread().getContextClassLoader();
try {
- this.sessionState = new SessionState(config.getHiveConf(),
+ // The session gets a conf of its own because it does not just read one:
its constructor
Review Comment:
🤖 nit: these inline comments are quite dense (7+ lines of narrative prose
here, similar blocks throughout). Could you tighten them to the key invariant —
e.g. "session needs its own HiveConf copy so close() doesn't delete scratch
dirs / close the loader shared with config's HiveConf" — and lean on the method
Javadocs for the rest? It'd be easier for a future reader to skim.
<sub><i>⚠️ AI-generated; verify before applying. React 👍/👎 to flag
quality.</i></sub>
--
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]
