search

grafana / helm-charts

Apache License 2.0
1.55k stars 2.21k forks source link

[Grafana] Incorrect dashboard visualization #3201

Open jpicara opened 2 weeks ago

jpicara commented 2 weeks ago

Hi, I would like to report the following unexpected issue we are suffering on different Grafana panels. Based on the different Kafka Prometheus metrics, we aim to visualize incoming/outgoing messages. Metric used: sum(irate(kafka_server_brokertopicmetrics_bytesin_total{}[1m])). On this way, we can see the following Prometheus graph for the last 7 days. imagen However, using the metric in Grafana, trend seems to be totally different to what we expect. imagen I would say this is related to the Max data points it is taking automatically, cause if I increase it from around 1500 that was taken automatically to random number just like 5000, output is clearly similar to what we got from Prometheus which is consistent with reality (at least peaks are quite similar) imagen How can we get the correct trend if this is related to the Max data points that is being taken automatically depends on the width of the panel? Thanks in advance for your appreciate support!

jpicara commented 1 week ago

Hello guys, Can anyone provide some light on this matter? Appreciate your support!

nwsteenberg commented 1 week ago

This is probably the wrong repo to take up this discussion. I would suggest you create the issue at https://github.com/grafana/grafana