If no data skew exists, you can set the job's parallelism any times of the count of taskmanagers, and set `cluster.evenly-spread-out-slots` to true in flink-conf.yaml of your flink cluster.
harshit.varsh...@iktara.ai <harshit.varsh...@iktara.ai> 于2022年11月7日周一 20:41写道: > > Dear Team, > > > > I need some advice on setting up load distribution of flink tasks. > > > > I have a flink task that processes transactions for users. Since load is more > than what can be handled on single machine, I want same task to be executed > on 3 machines. > > > > I am trying to use parallelism feature of flink for same. I am able to get > flink Job manager to start same task on 3 machines. I want task on each > machine to handle 1/3 of total user transaction, kindly suggest what > mechanism to use so each task handles only data for their own 1/3 of users. > > > > > > Thanks, > > Harshit Varshney