Hello, I would like to bring this to vote in the next day or two. Let me know if you have further comments.
Thanks, Allen On Thu, Mar 22, 2018 at 11:37 AM, Xavier Léauté <xav...@confluent.io> wrote: > > > > This kind of change will be problematic to us as the total RequestsPerSec > > will be double counted in our metric system as we do automatic > aggregation. > > It has been a headache for us as we always have to do some special > handling > > when querying such metrics. > > > > I agree with Allen, most of the metrics are already setup to be aggregated > across the various tags dimensions. We do the same on with our metrics, and > any respectable metrics system would allow you to do the same. > > It sounds like backwards compatibility in this case is more subjective, > since it depends largely on how you consume / aggregate those metrics, but > in my opinion, aggregation should be the way to go. > > There is precedent with topic based metrics such as BytesInPerSec, which > are currently duplicated, once at the broker level (no topic tag), and once > at the topic level (tagged with the topic name) and it's easy to miss that > fact. Without looking the source code it's very hard to notice those > subtleties. > > As a result it can be very error-prone to filter out the right metrics for > aggregation, so I would argue we should not repeat the same mistake unless > we had some good reasons to do so in the first place. >
