You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In reference to a Gitter discussion with @joe-elliott and this #425
Requirement - what kind of business use case are you trying to solve?
if an operation does not occur much and its a child of subsequent operations that are under a probabilistic sampling rate. Jaeger agent does not ensure that it would be represented a least once. (so we can draw the dependency graph).
Problem - what in Jaeger blocks you from solving the requirement?
At least 1 span should be captured from each unique operation to ensure completness if ann operation does not happen much but its a child of another operation that occurs lots. For example operation A (high freq, low sampling rate)=>operation B (low freq, high sampling rate) but it could be that the low sampling rate of operation A will also sample B making it very improbably to catch any operation b spans. In my use case, I am more interested in the completeness that are operated at least represented once. By completness I mean all each unique operation is represented at least once.
Proposal - what do you suggest to solve the problem or improve the existing situation?
In the scope of tail based sampling #425 ,we can make sure that we sample probabliscaly after a certain threshold, aka after catching a certain number of spans. It could be in the sampling config.
The text was updated successfully, but these errors were encountered:
In reference to a Gitter discussion with @joe-elliott and this #425
Requirement - what kind of business use case are you trying to solve?
if an operation does not occur much and its a child of subsequent operations that are under a probabilistic sampling rate. Jaeger agent does not ensure that it would be represented a least once. (so we can draw the dependency graph).
Problem - what in Jaeger blocks you from solving the requirement?
At least 1 span should be captured from each unique operation to ensure completness if ann operation does not happen much but its a child of another operation that occurs lots. For example operation A (high freq, low sampling rate)=>operation B (low freq, high sampling rate) but it could be that the low sampling rate of operation A will also sample B making it very improbably to catch any operation b spans. In my use case, I am more interested in the completeness that are operated at least represented once. By completness I mean all each unique operation is represented at least once.
Proposal - what do you suggest to solve the problem or improve the existing situation?
In the scope of tail based sampling #425 ,we can make sure that we sample probabliscaly after a certain threshold, aka after catching a certain number of spans. It could be in the sampling config.
The text was updated successfully, but these errors were encountered: