ElfHosted / Funky Penguin•5mo ago

Traefik not deploying

I have followed the cookbook and trying to get Traefik deployed but no pods are running under the traefik namespace. The only error I could find: kubectl logs -n flux-system deployment/kustomize-controller | grep -i error

"error":"Kustomization/flux-system/reflector dry-run failed: failed to prune fields: failed to convert merged object to last applied version: .spec.validation: field not declared in schema\nNamespace/traefik created\n"}

"error":"Kustomization/flux-system/reflector dry-run failed: failed to prune fields: failed to convert merged object to last applied version: .spec.validation: field not declared in schema\nNamespace/traefik created\n"}

configmap-traefik-he...

helmrelease-traefik....

helmrepository-traef...

kustomization-traefi...

namespace-traefik.ya...

12 Replies

TechHermitD•5mo ago

That sounds like a mal-formed configmap. On my phone right now so I can’t open the Yaml file but check your indentation.

Don Bish•5mo ago

I removed valuesFrom: from the helmrelease file to test having no config, but get same error if I include it or not

TechHermitD•5mo ago

Oh wait. That looks familiar, can you paste the top of the helm release here in text

Don Bish•5mo ago

apiVersion: helm.toolkit.fluxcd.io/v2beta1
kind: HelmRelease
metadata:
  name: traefik
  namespace: traefik
spec:
  chart:
    spec:
      chart: traefik
      version: 10.x 
      sourceRef:
        kind: HelmRepository
        name: traefik
        namespace: flux-system
  interval: 15m
  timeout: 5m
  releaseName: traefik
  valuesFrom:
  - kind: ConfigMap
    name: traefik-helm-chart-value-overrides
    valuesKey: values.yaml # This is the default, but best to be explicit for clarity

apiVersion: helm.toolkit.fluxcd.io/v2beta1
kind: HelmRelease
metadata:
  name: traefik
  namespace: traefik
spec:
  chart:
    spec:
      chart: traefik
      version: 10.x 
      sourceRef:
        kind: HelmRepository
        name: traefik
        namespace: flux-system
  interval: 15m
  timeout: 5m
  releaseName: traefik
  valuesFrom:
  - kind: ConfigMap
    name: traefik-helm-chart-value-overrides
    valuesKey: values.yaml # This is the default, but best to be explicit for clarity

TechHermitD•5mo ago

That looks right. There have been some issues in the past with the instructions on the site having a value that is no longer supported but I don’t see that in your config

root•5mo ago

can you link the instructions you followed? here are the ones I use (mine): https://github.com/faangbait/cluster/blob/main/README.md#configure-traefik and https://github.com/faangbait/cluster/blob/main/_init/10-storage.yaml and https://github.com/faangbait/cluster/blob/main/infrastructure/networking/traefik-values.yaml the thing that's failing is the prune: true line in your Kustomize. Try k delete ns traefik and then apply it again. crazy that somebody's actually recommending kustomize these days though.

Don Bish•5mo ago

I'm using the Geek cookbook https://geek-cookbook.funkypenguin.co.nz/kubernetes/ingress/traefik/

Why I use Traefik Ingress Controller - Funky Penguin's Geek Cookbook

Among other advantages, I no longer need to replicate SSL certificate secrets for nginx-ingress-controller to consume, once-per-namespace!

root•5mo ago

huh. how bout that. it's got gitopos now.

Don Bish•5mo ago

I managed to fix the traefik installation problem. Just reset all configs. @root is the cookbook using Kustomize not a good idea?

root•5mo ago

It's good to learn flux but man, I would space it out a little more than DP does. traefik is already like fifteen abstractions deeper than tcpdump. using helm to install it? okay, i'll buy that. but using flux to gitops it... i'd just draw the line before that is all. the more complicated things get, the more prone to failure they are. but he's the guy with a 2300 person discord server and i'm just a joker with a keyboard

Don Bish•5mo ago

Thanks for the input. I was installing everything via helm manually but I like the idea of doing it once in a repo and then I can replciated my setup should I make a new cluster in the future.

root•5mo ago

back in my day, we made things repeatable by writing documentation. if i've learned anything it's that you will never successfully one-click install this into a new cluster after it sits around unmaintained for two years. with documentation, you can see exactly which steps break and fix them as you go. as opposed to some fatal error deep in one of fifty places log lines live.

Gaming

Programming

Traefik not deploying