Wrong Metrics being published

[ghost]

I installed the library with all the supports in one of the production machines we have. This machine on an average serves about 10-13 request per second. But when I see this data in grafana it always overestimates. It starts from 12-13 but goes till 360-1k request per second. I tried to querry unerlying data in prometheus.

This is what I got for one particualr endpoint:

I might be wrong but shouldn't this number be ever increasing. Is there something wrong with my interpretation over here.

I am polling data every 30 seconds. I have not made any changes in the code base.

Please help. Let me know if I need to add any more details.