[GitHub] [flink] zentol commented on a change in pull request #10082: [FLINK-14164][runtime] Add a meter ‘numberOfRestarts’ to show number of restarts as well as its rate

[GitBox <gi...@apache.org>]

zentol commented on a change in pull request #10082: [FLINK-14164][runtime] Add a meter ‘numberOfRestarts’ to show number of restarts as well as its rate
URL: https://github.com/apache/flink/pull/10082#discussion_r343059023
 
 

 ##########
 File path: flink-runtime/src/main/java/org/apache/flink/runtime/scheduler/SchedulerBase.java
 ##########
 @@ -193,6 +197,11 @@ public SchedulerBase(
 		this.failoverTopology = executionGraph.getFailoverTopology();
 
 		this.inputsLocationsRetriever = new ExecutionGraphToInputsLocationsRetrieverAdapter(executionGraph);
+
+		// Use the counter from execution graph to avoid modifying execution graph interfaces
+		// Can be a new SimpleCounter created here after the legacy scheduler is removed.
+		this.numberOfRestartsCounter = executionGraph.getNumberOfRestartsCounter();
+		jobManagerJobMetricGroup.meter(NUMBER_OF_RESTARTS, new MeterView(numberOfRestartsCounter));
 
 Review comment:
   I agree that the meter doesn't make sense; the volume of events is simply to low.
   
   You'd want something it to work on hours I suppose, which is possible to implement (it's just a slightly modified MeterView, see below for the differences), but we'd technically violate the meter interface as it explicitly says that rates are per second.
   
   And I am wondering whether it wouldn't be perfectly fine to push this responsibility to the user, given that all time-series databases support queries that can derive the rate in whatever granularity they desire.
   That's my overall stance on metrics anyway; expose the basic raw data, and have users build more sophisticated things on top, with the only exception being things we want in the WebUI.
   
   
   
   How to implement a HourMeter:
   - take the `MeterView`
   - pin the values array to 60 values
   - don't divide by the timeSpan when calculating the rate
   - only act on every 12th update() call

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services