problem: ingress health check always be processing

[k8sedu]

when i use traefik as ingress controller, it whill not update status.loadbalancer field, so as to argo-cd, the ingress object will be processing.See traefik/traefik#3377, they will fix it.
can we change the method for ingress health check, or expose it as a config

[k8sedu]

when use traefik as ingress controller, field u check is as below:

  "status": {
    "loadBalancer": {}
  }

[alexmt]

Custom health checks are not yet supported but we plan to work on it in next release. #911

[k8sedu]

but
ty. could u add one config that it can be skipped since that different ingress controllers make different behaviors

[jessesuen]

@k8sedu according to traefik/traefik#3377, it seems the v1.7.X of Traefik (which is already released), supports this.

Since the current version of Traefik has the desired behavior, I don't see the benefit of adding this, especially since the custom health assessment feature would immediately obsolete the config setting we add.

[jessesuen]

For the record, Heptio's Contour ingress controller suffers the same problem:
projectcontour/contour#403

[k8sedu]

@k8sedu according to containous/traefik#3377, it seems the v1.7.X of Traefik (which is already released), supports this.

Since the current version of Traefik has the desired behavior, I don't see the benefit of adding this, especially since the custom health assessment feature would immediately obsolete the config setting we add.

my installed traefik 1.7.6 has not worked correctly. that field is empty.

[jessesuen]

@k8sedu this appears to be an option that needs to configured. See:

https://docs.traefik.io/configuration/backends/kubernetes/

I think you need to use publishedService field:

# Enable IngressEndpoint configuration.
# This will allow Traefik to update the status section of ingress objects, if desired.
#
# Optional
#
# [kubernetes.ingressEndpoint]
#
# At least one must be configured.
# `publishedservice` will override the `hostname` and `ip` settings if configured.
#
# hostname = "localhost"
# ip = "127.0.0.1"
# publishedService = "namespace/servicename"

[ermazi]

@jessesuen ty

[ermazi]

it worked.but there are some wrong with traefik doc about this.:D
official config omitting rbac update~

step1. add config

    [kubernetes]
      [kubernetes.ingressEndpoint]
        hostname = "traefik"

step2. update clusterrole

apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: traefik-ingress-controller
rules:
- apiGroups:
  - ""
  resources:
  - services
  - endpoints
  - secrets
  verbs:
  - get
  - list
  - watch
- apiGroups:
  - extensions
  resources:
  - ingresses
  - ingresses/status
  verbs:
  - get
  - list
  - watch
  - update

[enys]

for anyone arriving here, the link is now : https://doc.traefik.io/traefik/providers/kubernetes-ingress/#publishedservice

[0dragosh]

If anyone encounters this with Traefik ingress controller deployed via Helm, here's the values.yaml to fix it, at least for v10.6.0:

providers:
  kubernetesIngress:
    publishedService:
      enabled: true

[mickkael]

it worked.but there are some wrong with traefik doc about this.:D official config omitting rbac update~

step1. add config

    [kubernetes]
      [kubernetes.ingressEndpoint]
        hostname = "traefik"

step2. update clusterrole

apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: traefik-ingress-controller
rules:
- apiGroups:
  - ""
  resources:
  - services
  - endpoints
  - secrets
  verbs:
  - get
  - list
  - watch
- apiGroups:
  - extensions
  resources:
  - ingresses
  - ingresses/status
  verbs:
  - get
  - list
  - watch
  - update

the apiGroups must now include networking.k8s.io

• apiGroups:
  • extensions
  • networking.k8s.io

[yacine-1]

for Traefik users, if the "publishedService" option didn't work, just add the following annotation to your deployment:
--providers.kubernetesingress.ingressendpoint.ip=<your LB ip>

[davisoares454]

For me, what worked was to put this as a new arg on my Traefik container on its deloyment:

--providers.kubernetesingress.ingressendpoint.publishedservice=traefik/traefik

It then updated every ingresses controlled by it with a loadbalancer status with the right value.