ingress-nginx response-duration default query is using the wrong metric
driv opened this issue · 0 comments
driv commented
Describe the bug
When using Nginx Ingress, the default query response-duration is using the wrong metric.
(rate(nginx_ingress_controller_ingress_upstream_latency_seconds_sum{ namespace="default", ingress="number-generator"}[1m])
/
rate(nginx_ingress_controller_ingress_upstream_latency_seconds_count{namespace="default", ingress="number-generator"}[1m])) * 1000
To Reproduce
Deploy a container that generates slow responses, the canary is promoted.
Expected behavior
This metric being used instead:
(rate(nginx_ingress_controller_response_duration_seconds_sum{}[1m])
/
rate(nginx_ingress_controller_response_duration_seconds_count{}[1m])) * 1000
Additional context
- Flagger version: 1.37.0
- Kubernetes version: 1.30
- Service Mesh provider: N/A
- Ingress provider: ingress-nginx