I reckon that one easy fix would be to configure the samples and sample age to be super high...
Or I could make the metrics fully deterministic with MockTime, but that seems a little overkill. I don't need to test that the stats themselves compute what they claim to compute. I am more attempting to verify: 1. I have in fact recorded the metrics, with the proper names 2. I recorded the correct values (i.e., I didn't use the current time in place of the latency, or a latency as a number of occurrences, etc.) 3. I *probably* selected the right stats (i.e., I didn't use a Total when I meant to use a Max, etc.) For these purposes, I think it's sufficient to: 1. check that the metrics are present and non-zero 2. check that the values are in the right ballpark, although when it's easy to see what the exact value should be, that's nicer. [ Full content available at: https://github.com/apache/kafka/pull/5742 ] This message was relayed via gitbox.apache.org for devnull@infra.apache.org
