Re: Any issues with reinterpretAsKeyedStream when scaling partitions?

> > My top guess was that the mapping of keys to task managers changes > depending on the operator. E.g. if the same key (regardless of the > operator) goes to the same task manager, then I'd assume it's fine to > support multiple inputs into an operator with reinterpretAsKeyedStream. I agree

Not cleanup Kubernetes Configmaps after execution success

Hi all, We have Flink jobs run on batch mode and get the job status via JobHandler. onJobExecuted

flink yarn-per-job???? ynm????????????

?? bin/flink run -ynm flink-test -t yarn-per-job --detached ./examples/streaming/TopSpeedWindowing.jar ??yarn ui??nameFlink per-job cluster

??????????

Hi ?? user-zh-unsubscr...@flink.apache.org Best JasonLee ??2021??10??25?? 10:52??zdj<1361776...@qq.com.INVALID> ??

????

Re: Any issues with reinterpretAsKeyedStream when scaling partitions?

On Sat, Oct 23, 2021 at 9:16 PM JING ZHANG wrote: > I found the following link about this. Still looks applicable. In my >> case, I don't need to do a broadcast join. >> >> https://www.alibabacloud.com/blog/flink-how-to-optimize-sql-performance-using-multiple-input-operators_597839 >> > This

Re: 在开启checkpoint后如何设置offset的自动提交以方便监控

Hi！ 这里的 offset 是 kafka source 的 offset 吗？其实没必要通过 checkpoint 读取 offset，可以通过 metrics 读取，见 [1]。 [1] https://ci.apache.org/projects/flink/flink-docs-master/docs/ops/metrics/#kafka-connectors 杨浩 于2021年10月25日周一 上午10:20写道： > 请问下，如果启用checkpoint，因为状态比较大，checkpoint间隔设置比较大，如何让offset提交的比较快，这样方便监控程序进度

Re: how to delete all rows one by one in batch execution mode;shutdown cluster after all tasks finished

Hi! Thanks for the clarification. Flink currently does not have the functionality to "revert all operations till some point". What I would suggest is still to discard the resulting tables and run the pipeline from the point when the filtering logic is changed. If the pipeline has processed some

在开启checkpoint后如何设置offset的自动提交以方便监控

请问下，如果启用checkpoint，因为状态比较大，checkpoint间隔设置比较大，如何让offset提交的比较快，这样方便监控程序进度