JkSelf commented on code in PR #11898:
URL: https://github.com/apache/gluten/pull/11898#discussion_r3058595786
##########
cpp/velox/substrait/SubstraitToVeloxPlan.cc:
##########
@@ -399,24 +399,27 @@ core::PlanNodePtr
SubstraitToVeloxPlanConverter::toVeloxPlan(const ::substrait::
} else if (
sJoin.has_advanced_extension() &&
SubstraitParser::configSetInOptimization(sJoin.advanced_extension(),
"isBHJ=")) {
+ auto isHashTableBuildOncePerExecutor =
+ SubstraitParser::configSetInOptimization(sJoin.advanced_extension(),
"isHashTableBuildOncePerExecutor=");
std::string hashTableId = sJoin.hashtableid();
std::shared_ptr<core::OpaqueHashTable> opaqueSharedHashTable = nullptr;
bool joinHasNullKeys = false;
- try {
- auto hashTableBuilder =
ObjectStore::retrieve<gluten::HashTableBuilder>(getJoin(hashTableId));
- joinHasNullKeys = hashTableBuilder->joinHasNullKeys();
- auto originalShared = hashTableBuilder->hashTable();
- opaqueSharedHashTable = std::shared_ptr<core::OpaqueHashTable>(
- originalShared,
reinterpret_cast<core::OpaqueHashTable*>(originalShared.get()));
-
- LOG(INFO) << "Successfully retrieved and aliased HashTable for reuse.
ID: " << hashTableId;
- } catch (const std::exception& e) {
- LOG(WARNING)
- << "Error retrieving HashTable from ObjectStore: " << e.what()
- << ". Falling back to building new table. To ensure correct results,
please verify that spark.gluten.velox.buildHashTableOncePerExecutor.enabled is
set to false.";
- opaqueSharedHashTable = nullptr;
+ if (isHashTableBuildOncePerExecutor) {
+ try {
+ auto hashTableBuilder =
ObjectStore::retrieve<gluten::HashTableBuilder>(getJoin(hashTableId));
+ joinHasNullKeys = hashTableBuilder->joinHasNullKeys();
+ auto originalShared = hashTableBuilder->hashTable();
+ opaqueSharedHashTable = std::shared_ptr<core::OpaqueHashTable>(
+ originalShared,
reinterpret_cast<core::OpaqueHashTable*>(originalShared.get()));
+
+ LOG(INFO) << "Successfully retrieved and aliased HashTable for reuse.
ID: " << hashTableId;
+ } catch (const std::exception& e) {
+ LOG(ERROR)
Review Comment:
When this configuration is enabled, the hash table must be retrieved from
the ObjectStore. If retrieval fails, an exception should be thrown on the
Gluten side rather than letting it propagate to the Velox side, consistent with
the approach in [PR #11749](https://github.com/apache/gluten/pull/11749). If
configuration is disabled, we already set opaqueSharedHashTable to null and no
need further process.
--
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
