apache
[Improvement] In the CatalogConnectorManager of Trino-Connector, when a single-threaded scan of the catalog encounters an exception, it does not exit and thus causes subsequent creation actions to keep timing out
What would you like to be improved?
The parameter NUMBER_EXECUTOR_THREAD of the CatalogConnectorManager class is fixed at 1. This is a single-threaded approach. Then in Trino, a catalog named gt_hive is created using the call command. When I made a creation request without providing all the parameters and received an error, an exception occurred. After that, the background log continuously showed that this thread was polling for the creation error. The log is as follows: 2025-11-13 11:19:08 ERROR [Query-20251113_031908_00312_nkzn5-371] CatalogConnectorManager:438 - Failed to create connector: gt_hive 2025-11-13 11:19:08 WARN [gravitino-connector-schedule-0] CatalogRegister:314 - Execute command failed: CREATE CATALOG gt_hive USING gravitino WITH .....
2025-11-13 11:19:13 ERROR [Query-20251113_031913_00313_nkzn5-354] CatalogConnectorManager:438 - Failed to create connector: gt_hive 2025-11-13 11:19:13 WARN [gravitino-connector-schedule-0] CatalogRegister:314 - Execute command failed: CREATE CATALOG gt_hive USING gravitino WITH .....
This log will be output every 5 seconds on average.
This will cause the subsequent creation of the catalog to keep timing out. Therefore, it is necessary to manually delete the abnormal catalog in Gravitino and restart Trino. Only then will the creation of the catalog in Trino succeed.
How should we improve?
No response
