Re: Inquiry about Possible Deadlock Problem in Hive 4.1.0 with Insert overwrite Query on Iceberg Table

2025-03-26 Thread 박근형
Thank you for your response. I understand that even if Hive and Trino use the same Hive Metastore as a Catalog, additional implementation in TrinoCatalog is necessary to ensure data correctness. To clarify, the deadlock issue I am referring to is not a situation where the IOW lock cannot be obtain

Re: Inquiry about Possible Deadlock Problem in Hive 4.1.0 with Insert overwrite Query on Iceberg Table

2025-03-26 Thread Denys Kuzmenko
Hi, What do you mean by "possible deadlock issue"? - IOW is stuck trying to acquire a write lock (could you check this in HS2 logs)? - What is the value of `hive.txn.ext.locking.enabled` config? If 'false', pessimistic locking won't be enabled on iceberg tables. - Are you using any Hive-mana

Re: Inquiry about Possible Deadlock Problem in Hive 4.1.0 with Insert overwrite Query on Iceberg Table

2025-03-26 Thread Butao Zhang
I have not checked the detail about HVIE-28366. Maybe Denys can give more explanations. But for your case: For iceberg table, IOW in Hive and Insert in Trino at the same time. If you want to ensure data correctness, You should ensure that Hive/Hs2 and Trino use the same Iceberg catalog. Howev

Re: Inquiry about Possible Deadlock Problem in Hive 4.1.0 with Insert overwrite Query on Iceberg Table

2025-03-26 Thread 박근형
Thank you for your response. Based on HIVE-28366, it appears that data correctness can be ensured through HMS Lock when 'Insert' and 'IOW' operations are executed concurrently. If both operations are performed by Hive, my understanding is that the problem is resolved regardless of the 'hive.txn.man