Average response time on Grafana from Prometheus

Question

I am trying to find the average response time of all the URLs in Grafana,

The below query not returning any data, what am I doing wrong here?

    avg by (uri) (rate(http_server_requests_seconds_sum{app_name="$app",
                  namespace="$namespace", uri!~".*actuator.*|/health|root"}[1m]) 
   /rate(http_server_requests_seconds_count{app_name="$app",
                  namespace="$namespace", uri!~".*actuator.*|/health|root"}[1m]))

The data is collected via a micrometer.

You aren't getting any result because the first part is eliminating all labels except the uri, while the dividend is keeping all the labels. Since there isn't a perfect match of labels anymore, Prometheus discards them. @pulak's solution below with using sum fixes that — checketts

Pulak Kanti Bhowmick Pulak Kanti Bhowmick · Accepted Answer · 2021-03-18T09:06:01

If you want to get the average then you should divide the _sum by the _count counter. If your labels are ok, then you will get data(if there any) by below PromQL.

sum by (uri) (rate(http_server_requests_seconds_sum{app_name="$app", 
            namespace="$namespace", uri!~".*actuator.*|/health|root"}[1m])) 
/ sum by (uri) (rate(http_server_requests_seconds_count{app_name="$app",
            namespace="$namespace", uri!~".*actuator.*|/health|root"}[1m]))

Average response time on Grafana from Prometheus

1 Answers