0

I try to implement tracing for my system. I use OpenCensus for tracing and Prometheus for metric backend. In OpenCensus, when defining a view, we will define Aggregation function. (i.e: count, last value, distribution ...)

My problem is: I have a metric, for example named cache_missed to express total cached miss when querying to caching server. This number definitely will increase overtime (and not have upper limit). I want to track data so when looking into dashboard, I will know there is a day that cache miss happen more often. In other word, I want to view the correlation between cache miss at different timestamp.

I think I can do that by using distribution for aggregation function. But I don't sure does it is possible and how can I achieve that. Please tell me.

Thanks

Trần Kim Dự
  • 5,872
  • 12
  • 55
  • 107
  • Do you mean you are looking for the [increase](https://prometheus.io/docs/prometheus/latest/querying/functions/#increase) in your metric ? Or is your question about having the data aligned on day boundaries ? – Michael Doubez Jun 02 '19 at 10:53
  • @MichaelDoubez thanks. Look like "increase" is the function I need :D – Trần Kim Dự Jun 02 '19 at 18:48
  • I am glad. Be wary of [aliasing issues](https://stackoverflow.com/questions/48218950/increase-in-prometheus-sometimes-doubles-values-how-to-avoid) – Michael Doubez Jun 03 '19 at 13:30

0 Answers0