Thanks for your reply! I tried to use avg_over_time and quantile_over_time, these 2 functions will still return a time series data based on the cpu_utilization data. So if I apply "< 0.10" on it, it will still return the data points which are < 0.1. This 2 functions are helpful, but still I didn't get the result I need. For example, I apply these 2 functions for the cpu utilization data on 4/1/2021, I still get a time series for it. How can I determine if the "cpu utilization on 4/1/2021 is above or below 20%"?
On Saturday, April 3, 2021 at 1:00:51 AM UTC-7 [email protected] wrote: > You want to use one of the "over time" functions. > > avg_over_time(cpu_utilization[1d]) < 0.10 > > quantile_over_time(0.9 cpu_utilization[1d]) < 0.10 > > On Sat, Apr 3, 2021 at 1:34 AM chuanjia xing <[email protected]> wrote: > >> Hi there, >> I think my question is pretty common but looks like I can't find the >> answer to it. >> The problem I am trying to solve is: Let's say I have 100 metrics >> data for cpuutilization for my instance. Each one is a time series for 1 >> day with granularity of 1 min. So each one has 1,440 data point. >> Now I want to select the instances which has an "average cpu >> utilization" <= 10% for one day. How should I do this, basically how should >> I calculate the "average cpu utilization" for one time series? >> One way I am thinking to do is, for each time series, I find the >> p90 data point and check if it is <= 10%. If it is, then I'll say that this >> instance data is with "average cpu utilization" <= 10% and select it. But >> looks like there's no function in Prometheus for me to find the p90 for a >> time series, or I just didn't find the right function for it. >> So, is above the right way to select? If yes, how can I find the >> p90 for a time series? If not, then how should I solve this problem? >> >> Thanks, >> >> -- >> You received this message because you are subscribed to the Google Groups >> "Prometheus Users" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to [email protected]. >> To view this discussion on the web visit >> https://groups.google.com/d/msgid/prometheus-users/89dfc6ac-2c05-4fab-8c04-eb68e32af214n%40googlegroups.com >> >> <https://groups.google.com/d/msgid/prometheus-users/89dfc6ac-2c05-4fab-8c04-eb68e32af214n%40googlegroups.com?utm_medium=email&utm_source=footer> >> . >> > -- You received this message because you are subscribed to the Google Groups "Prometheus Users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/prometheus-users/9a1f8455-362f-4b27-8ca6-913636d52f59n%40googlegroups.com.
