actor_execution_time and actor_poll_duration both use total_poll_duration

[lmatz]

Describe the bug

https://github.com/risingwavelabs/risingwave/blob/main/src/stream/src/task/stream_manager.rs#L719-L722

which value is it supposed to use? didn't find it

https://github.com/risingwavelabs/risingwave/blob/main/src/stream/src/task/stream_manager.rs#L739-L742

Additional context

I was looking for a metric that can show the processing time (avg, 99th, etc) of one row in each operator, is this actor_execution_time expected to record this? to detect all the bottlenecks of a stream query

(per row probably generates too much data points, sampling or per chunk is also good enough)

correct me if I am wrong, but the back pressure metric/graph shows only a portion of bottlenecks, doesn't it?
e.g. both upstream and downstream operators are slow and slow to the same extent

I think it could be a good first metric when investigating the scaling-up/out performance.

Is it also possible to show it on RW's dashboard together with the fragment graph

[github-actions]

This issue has been open for 60 days with no activity. Could you please update the status? Feel free to continue discussion or close as not planned.