Skip to content

Latest commit

 

History

History
180 lines (116 loc) · 7.63 KB

CHANGES.md

File metadata and controls

180 lines (116 loc) · 7.63 KB

caktus.k8s-web-cluster

This project uses semantic versioning.

Changes

v1.7.0 on December 5th, 2024

  • Use boolean type for installCRDs cert-manager value to support v1.15.0+. Eventually, this value should be migrated to crds.keep: true and crds.enabled: true.

v1.6.0 on May 8th, 2024

v1.5.0 on April 11th, 2023

  • Requires Ansible v6.0+
  • Switch to kubernetes.core for Ansible 6.x+ support. The community.kubernetes collection was renamed to kubernetes.core in v2.0.0 of the kubernetes.core collection. Since Ansible v3.0.0, both the kubernetes.core and community.kubernetes namespaced collections were included for convenience. Ansible v6.0.0 removed the community.kubernetes convenience package.
  • Use fully qualified collection names (FQCNs) to be explicit
  • Add k8s_cert_manager_release_values variable to allow per-project customization of Helm chart values

v1.4.0 on February 8th, 2023

  • Add optional Descheduler for Kubernetes support. Enable with k8s_install_descheduler and reference defaults/main.yml for configuration options.

v1.3.0 on April 20th, 2022

  • Set nginx podAntiAffinity to default to be a nice default that prefers (but does not require) scheduling pods on different nodes. Override with k8s_ingress_nginx_affinity.
  • Allow configuration of nginx service loadBalancerIP via k8s_ingress_nginx_load_balancer_ip.
  • Add extendable k8s_cert_manager_solvers variable to support configuring DNS01 challenge provider
  • Update api version for echotest ingress

v1.2.0 on Dec 6th 2021

  • Default to 2 replicas for ingress controller
  • Fix typo in k8s_digitalocean_load_balancer_hostname

v1.1.0 on Mar 12, 2021

  • Move CI user creation to caktus.django-k8s role since it is something that is project or environment-specific.

v1.0.0 on Feb 18, 2021

BACKWARDS INCOMPATIBLE CHANGES:

  • Use Helm to install ingress-nginx
    • ingress-nginx controller upgraded from 0.26.1 to 0.44.0 (via 3.23.0 helm chart release)
  • Use Helm to install cert-manager
    • cert-manager controller upgraded from v0.10.1 to v1.2.0
    • The accompanying caktus.django-k8s role must also be updated to >v0.0.11 to restore certificate validation.
  • You must follow the Digital Ocean instructions and set a hostname via k8s_digitalocean_loadbalancer_hostname to keep PROXY protocol enabled on Digital Ocean (required to see real client IP addresses).

Upgrade instructions:

  1. First, purge the old cert-manager and create a new ingress controller in a new namespace:

    # Install new ingress controller, but don't delete the old one yet
k8s_install_ingress_controller: yes
k8s_ingress_nginx_namespace: ingress-nginx-temp
k8s_purge_ingress_controller: no

# Don't install a new cert-manager, but do delete the old one
k8s_install_cert_manager: no
k8s_purge_cert_manager: yes

    If you don't wish to make two DNS changes, you may find it helpful to set k8s_ingress_nginx_namespace to a more permanent name.

    $ ansible-playbook -l <host/group> deploy.yaml -vv

  2. Look up the IP or hostname for the new ingress controller:

    $ kubectl -n ingress-nginx-temp get svc

  3. Change the DNS for all domains that point to this cluster to use the new IP or hostname. You may find it helpful to watch the logs of both ingress controllers during this time to see the traffic switch to the new ingress controller.

    The post Kubernetes: Nginx and Zero Downtime in Production has a more detailed overview of this approach.

  4. Next, add k8s_purge_ingress_controller: yes to your variables file and re-run deploy.yaml. Note that you will now have both k8s_install_ingress_controller: yes and k8s_purge_ingress_controller: yes, however, the former refers to the new namespace and the latter refers only to the old namespace. This should clear out the old ingress controller.

    Note, you may need to run this a few times if Ansible times out attempting to delete everything the first time.

  5. If you want to switch everything to use the original ingress-nginx namespace again, make the change in your variables file and re-run deploy.yaml with your final configuration.

    Otherwise, simply set k8s_install_cert_manager: yes and do not change the namespace.

    # your variables file (e.g., group_vars/all.yaml)
k8s_install_ingress_controller: yes
k8s_ingress_nginx_namespace: ingress-nginx

k8s_install_cert_manager: yes

    Make sure to remove the two k8s_purge_* variables as they are no longer needed and will be removed in a future release.

  6. If you elected to switch namespaces again:

    • Change the DNS to the new service address as in step 3 and wait for traffic to stop going to the temporary ingress controller.

    • Remove the ingress-nginx-temp namespace as follows:

      helm -n ingress-nginx-temp uninstall ingress-nginx
kubectl delete ns ingress-nginx-temp

  7. Test that cert-manager is working properly by deploying a new echotest pod as described in the README.

  8. Update any projects that deploy to the cluster to use the corresponding 1.0 release of ansible-role-django-k8s.

Please note that the k8s_purge_* variables are intended only for removing the previously-installed versions of these resources. If you need to remove the newly installed cert-manager or ingress-nginx for any reason, you should use the helm uninstall method described above.

Other Changes:

  • Move Papertrail and New Relic to caktus.k8s-hosting-services. The existing deployments will not be automatically removed, but they are no longer managed from this role. To take advantage of future changes to those deployments, add the caktus.k8s-hosting-services role to your requirements.yaml file
  • Retire k8s_cluster_name variable.

v0.0.7 on Jul 6, 2020

  • Allow Papertrail memory resources to be configurable

v0.0.6 on Jul 2, 2020

  • Support creation of an AWS IAM user with limited perms that can be used on CI to push images and deploy.

v0.0.5 on Jun 29, 2020

  • Introduce k8s_cluster_name variable

v0.0.4 on Jun 28, 2020

  • On AWS, grant cluster access to IAM users in k8s_iam_users.

v0.0.3 on Jun 25, 2020

  • Re-enable validation in cert-manager workspace after installing or updating cert-manager and Lets Encrypt.
  • Add NewRelic Infrastructure support

v0.0.2 on Mar 1, 2020

  • Added Logspout for Papertrail

v0.0.1 on Feb 26, 2020

  • Initial release