Disable Aggregagation Based on Time Range

Grafana Prometheus
1

I suspect based on previous posts there is no way to do this but thought to ask one more time.

When i visualize cpu util in a panel (chart) over a 1 hour period, i get a data point every 15 seconds
When i switch to 24 hours, i get a data point every 1 minute
When i switch to 90 days, i get a data point every 1 hour

Is there any way to stop this such that at 90 days i get every 15 seconds? I understand the issue with visualizing this much data so having it in a chart is not 100% necessary, but table visualization does the same.

also, can someone help explain when it does aggregate, what number am i seeing (max for that timeframe, average for that timeframe)?

2

That’s 0.5+M data points. I hope you know what you’re doing :smiley:

Play with Min interval and Max data points config options to achieve your needs:

obrázok
obrázok2895×1050 248 KB

3

ive done that and i get the error

exceeded maximum resolution of 11,000 points per timeseries. Try decreasing the query resolution (?step=XX)

which i beleive is actually prometheus?

if this isnt possible, is there a way to determine what number is actually being returned when the step “auto” sets to 1h?

4

Yes. Developers know that users can kill their own Prometheus and they complain that it doesn’t work - so I guess this is some safety limit.

Use Query inspector to find executed query values:

obrázok
obrázok1476×1843 272 KB