Service profiles and nginx ingress

[schwichti]

Hi,

I am using Linkerd edge-21.9.3 in conjunction with nginx ingress controller. I want to configure service profiles for retries and timeouts.
Sidecars are up and running.
As described at https://linkerd.io/2.10/tasks/using-ingress/#nginx, I have added the annotation to my Ingresses:
nginx.ingress.kubernetes.io/service-upstream: true.

The service profile looks like this (values are for debugging):

apiVersion: linkerd.io/v1alpha2
kind: ServiceProfile
metadata:
  name: appshell.simphera.svc.cluster.local
  namespace: simphera
spec:
  routes:
  - name: 'GET /projects'
    condition:
      pathRegex: '/projects'
      method: GET
    isRetryable: true
    timeout: 60000ms
  retryBudget:
    retryRatio: 1.0
    minRetriesPerSecond: 10
    ttl: 10s

Retries does not seem to work and when I run linkerd viz routes this is what I see:

ROUTE            SERVICE   SUCCESS   RPS   LATENCY_P50   LATENCY_P95   LATENCY_P99
GET /projects   appshell         -     -             -             -             -
[DEFAULT]       appshell         -     -             -             -             -

Any hint what is going wrong?

[adleong]

Hi @schwichti! Just to clarify, you're looking to configure retries and timeouts for the calls from the ingress controller to your application, right? Am I correct in assuming that appshell is your application's service?

Can you also clarify the full arguments you are passing to linkerd viz routes? Retries and timeouts are performed on the client side so in order to see them, you need to look at things from the ingress controller's point-of-view. i.e. linkerd viz routes <ingress controller> --to <application> -o wide. For more details see: https://linkerd.io/2.10/tasks/configuring-retries/#monitoring-retries