Details
-
Improvement
-
Status: Open
-
Not a Priority
-
Resolution: Unresolved
-
None
-
None
Description
Now, flink has provided latency metrics to monitor the latency, but this function mainly used in debugging contexts rather than in production contexts because of throughput effect. If we provider an api that can dynamic open/close this function, then we can monitor the online data latency and find out the performance bottleneck in time without restart the job, which maybe helpful for online debug.Â