Thanks for feedback! Can you post the jobmanager log? This may be because the source parallelism has been set in some way (Kind of source can infer parallelism according to the catalog).
Best, Lijie Edwin <xuhao...@163.com> 于2022年3月1日周二 19:30写道： > Hi all, > > I was trying to run flink-tpcds tests on flink v1.15-snapshot, and I added > "jobmanager.adaptive-batch-scheduler.default-source-parallelism: 8" along > with some other configurations to the conf yaml file to enable adaptive job > scheduler. But it turned out the source parallelism kept unchanged(always > be 1), I was wondering if anyone could kindly give some advice on how to > fix this? some of my configurations are listed below: > > jobmanager.scheduler: AdaptiveBatch > jobmanager.adaptive-batch-scheduler.max-parallelism: 64 > jobmanager.adaptive-batch-scheduler.data-volume-per-task: 6m > jobmanager.adaptive-batch-scheduler.default-source-parallelism: 8 > > metrics.reporter.promgateway.class: > org.apache.flink.metrics.prometheus.PrometheusPushGatewayReporter > metrics.reporter.promgateway.randomJobNameSuffix: true > metrics.reporter.promgateway.deleteOnShutdown: true > > parallelism.default: -1 > > Thanks! > > > > >
