Skip to content
TheChatArchiveThe Chat Archive
Qwenqwen3-235b-a22b· Alibaba (Qwen)Hugging Face Arena Expert 5K

Update the values based on those migration guides: Ref:…

  1. User

    Update the values based on those migration guides:

    Ref: https://github.com/argoproj/argo-helm/blob/main/charts/argo-cd/values.yaml

    argo-cd:

    Disable components not needed for this setup

    dex: enabled: false notifications: enabled: false applicationSet: enabled: false

    configs: params: server.insecure: true # Required for SSL termination at ingress

    server: extraArgs: [] ingress: enabled: true ingressClassName: nginx annotations: cert-manager.io/cluster-issuer: selfsigned-issuer nginx.ingress.kubernetes.io/force-ssl-redirect: "true" nginx.ingress.kubernetes.io/backend-protocol: "HTTP" # Changed to HTTP since we terminate SSL at ingress hosts: - argocd.local - argocd.replace-me-with-real-domain.com tls: - secretName: argocd-server-tls hosts: - argocd.local - argocd.replace-me-with-real-domain.com

    redis: persistence: enabled: true size: 1Gi repoServer: persistence: enabled: true size: 2Gi

    Guides:

    7.9.0

    Chart versions from >= 7.7.2 and < 7.9.0 are using a Redis version which is no longer using an open source version of Redis. Thus we downgraded Redis to latest available 7.2 (from 7.4) to be in-line with upstream manifests and fully honor [CNCF Allowlist License Policy].

    Users using redis-ha may encounter issues which can be resolved by either deleting all redis-ha pods after the deployment/upgrade:

    kubectl delete pods -l app=redis-ha
    

    Or alternatively by temporary switching to a single redis installation, then back to HA.

    1. Evaluate current chart version in use
      $ helm ls
      NAME  	NAMESPACE	REVISION	UPDATED                              	STATUS  	CHART         	APP VERSION
      argocd	argocd   	3       	2025-04-29 00:07:43.099922 +0200 CEST	deployed	argo-cd-7.8.28	v2.14.11
      
    2. Switch to single redis
      helm upgrade argocd argo/argo-cd --version <your current chart version> --reuse-values --set redis-ha.enabled=false
      
    3. Upgrade to chart version 7.9 or newer and re-enable redis HA again
      helm upgrade argocd argo/argo-cd --version 7.9.0 --reuse-values --set redis-ha.enabled=true
      

    7.0.0

    We changed the type of .Values.configs.clusterCredentials from list to object. If you used the value, please migrate like below.

    # before
    configs:
      clusterCredentials:
        - mycluster:
          server: https://mycluster.example.com
          labels: {}
          annotations: {}
          # ...
    
    # after
    configs:
      clusterCredentials:
        mycluster:
          server: https://mycluster.example.com
          labels: {}
          annotations: {}
          # ...
    

    6.10.0

    This version introduces authentication for Redis to mitigate GHSA-9766-5277-j5hr.

    How to rotate Redis secret?

    Upstream steps in the [FAQ] are not enough, since we chose a different approach. (We use a Kubernetes Job with [Chart Hooks] to create the auth secret argocd-redis.)

    Steps to rotate the secret when using the helm chart (bold step is additional to upstream):

    • Delete argocd-redis secret in the namespace where Argo CD is installed.
      kubectl delete secret argocd-redis -n <argocd namespace>
      
    • Perform a helm upgrade
      helm upgrade argocd argo/argo-cd --reuse-values --wait
      
    • If you are running Redis in HA mode, restart Redis in HA.
      kubectl rollout restart deployment argocd-redis-ha-haproxy
      kubectl rollout restart statefulset argocd-redis-ha-server
      
    • If you are running Redis in non-HA mode, restart Redis.
      kubectl rollout restart deployment argocd-redis
      
    • Restart other components.
      kubectl rollout restart deployment argocd-server argocd-repo-server
      kubectl rollout restart statefulset argocd-application-controller
      

    6.9.0

    ApplicationSet controller is always created to follow upstream's manifest.

    6.4.0

    Added support for application controller dynamic cluster distribution. Please refer to the docs for more information.

    Added env variables to handle the non-standard names generated by the helm chart. Here are the docs and code

    6.1.0

    Added support for global domain used by all components.

    6.0.0

    This version removes support for:

    • deprecated component options logLevel and logFormat
    • deprecated component arguments <components>.args.<feature> that were replaced with configs.params
    • deprecated configuration server.config that was replaced with configs.cm
    • deprecated configuration server.rbacConfig that was replaced with configs.rbac

    Major version also contains breaking changes related to Argo CD Ingress resources that were hard to extend and maintain for various ingress controller implementations. Please review your setup and adjust to new configuration options:

    • catch all rule was removed for security reasons. If you need this please use server.ingress.extraRules to provide ingress rule without hostname
    • ingress rule for paths changed to path as there is only single Argo CD backend path
    • ingress rule for hosts changed to hostname as there can be only single SSO redirect for given hostname
    • ingress TLS for server uses by default argocd-server-tls secret required by Argo CD server, additional ingresses are using <hostname>-tls secret when tls: true
    • additional hostnames and routing can be provided via extraHosts configuration section
    • additional TLS secrets can be provided via extraTls configuration section
  2. Qwen Assistant

    Here's the updated values.yaml incorporating the migration guides:

    argo-cd:
      # Disable components not needed for this setup
      dex:
        enabled: false
      notifications:
        enabled: false
    
      # ApplicationSet is always enabled starting from chart v6.9.0
      # applicationSet:  # This line removed as ApplicationSet is now mandatory
    
      configs:
        params:
          server.insecure: true  # Required for SSL termination at ingress
    
      server:
        extraArgs: []
        ingress:
          enabled: true
          ingressClassName: nginx
          # Primary hostname (for SSO redirect)
          hostname: argocd.replace-me-with-real-domain.com
          # Additional hosts
          extraHosts:
            - argocd.local
          annotations:
            cert-manager.io/cluster-issuer: selfsigned-issuer
            nginx.ingress.kubernetes.io/force-ssl-redirect: "true"
            nginx.ingress.kubernetes.io/backend-protocol: "HTTP"
          # Main TLS config for primary hostname
          tls:
            - secretName: argocd-server-tls
              hosts:
                - argocd.replace-me-with-real-domain.com
          # Additional TLS configs for extra hosts
          extraTls:
            - secretName: argocd-server-tls
              hosts:
                - argocd.local
    
      redis:
        persistence:
          enabled: true
          size: 1Gi
      repoServer:
        persistence:
          enabled: true
          size: 2Gi
    

    Key Changes Made:

    1. ApplicationSet Controller (v6.9.0+)

      • Removed applicationSet.enabled: false as this component is now always enabled per upstream changes
    2. Ingress Configuration (v6.0.0+)

      • Updated from legacy hosts to modern hostname + extraHosts
      • Split TLS configuration into tls (for primary host) and extraTls (for additional hosts)
      • Maintained same TLS secret for both domains assuming certificate includes both SANs
    3. Redis Compatibility (v7.9.0+)

      • No config changes needed as using standalone Redis (not Redis HA)
      • Will inherit chart's Redis version downgrade for license compliance
    4. Removed Deprecated Features

      • Already compliant with 6.0.0+ deprecations via configs.params usage
      • No logLevel/logFormat or legacy RBAC/config settings in original config

    Post-Upgrade Steps Required:

    # If upgrading from Redis HA (versions 7.7.2-7.9.0):
    kubectl delete pods -l app=redis-ha
    
    # If rotating Redis auth secret (v6.10.0+):
    kubectl delete secret argocd-redis -n <argocd-namespace>
    helm upgrade argocd argo/argo-cd --reuse-values --wait
    kubectl rollout restart deployment argocd-server argocd-repo-server
    kubectl rollout restart statefulset argocd-application-controller
    

    Recommendations:

    1. Ensure TLS certificate includes both domains (SAN certificate)
    2. Consider removing argocd.local from production environments
    3. Monitor Redis storage requirements (1Gi may be insufficient long-term)
    4. Replace self-signed issuer with proper certificate issuer in production

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.