kangTwang created FLINK-26579: --------------------------------- Summary: The default source parallelism configuration in adaptive batch scheduling does not take effect Key: FLINK-26579 URL: https://issues.apache.org/jira/browse/FLINK-26579 Project: Flink Issue Type: Bug Components: API / DataStream Affects Versions: 1.15.0 Reporter: kangTwang Fix For: 1.15.0
In the adaptive batch scheduler, the parameter "jobmanager. Adaptive-batch-scheduler. Default source parallelism" configured in the tpc-ds test does not take effect, resulting in that only one record will be received in the downstream. Even if the csvtablesource has 64 parallelism, it will only be executed with one parallelism, resulting in poor performance of 99 SQL in the data reading stage -- This message was sent by Atlassian Jira (v8.20.1#820001)