how to implement multi-thread source generate?

[varuy322]

Hi, there
1)Could you give some advice that how to implement multiply thread to generate source data to Flink?
2)It didn't print the latency value into .out file under taskmanager log, but Throughput info existed. The code couldn't jump into the branch which prevTime not equal to -1. What's more, the WindowedCount value could sink into kafka. I think it will be more convenient to read throughput and latency in local file directly rather than to read from kafka.
Someone may give some suggestion for me? Thanks a ton!

[grtheod]

Hi,

1. You could just change the parallelism and adjust properly the number of records per second.
2. If you redirect worker's output to the .out file, everything will be there. The latency measurement is not ideal, I used more like an estimation, so maybe it would be better to find an alternative way (e.g., use Kafka). In the current version, Kafka is not utilized for Flink and we manually use logs.