Skip to content

CHANGELOG.md

Latest commit

 

History

History
227 lines (142 loc) · 10.2 KB

CHANGELOG.md

File metadata and controls

227 lines (142 loc) · 10.2 KB

Changelog

  • BREAKING

    • CNI bin_dir in CRI runtime config is deprecated (plugins.'io.containerd.cri.v1.runtime'.cni.bin_dir) and will be removed in containerd v2.3. It was replaced with bin_dirs in the same section which supports a list of directories. So, plugins.'io.containerd.cri.v1.runtime'.cni.bin_dir = '/opt/cni/bin' was changed to plugins.'io.containerd.cri.v1.runtime'.cni.bin_dirs = ['/opt/cni/bin'] in containerd_config variable.

  • UPDATE

    • update containerd to v2.2.1

  • FEATURE

  • MOLECULE

    • add test for conf.d include feature

0.16.0+2.1.4

  • Breaking

    • meta/main.yml: Change min_ansible_version to 2.15. Ansible 2.9 is end-of-life (EOL).

  • UPDATE

    • update containerd to v2.1.4
    • update .gitignore
    • update .yamllint
    • fix ansible-lint issues

0.15.0+2.1.3

  • UPDATE

    • update containerd to v2.1.3

  • MOLECULE

    • Use generic/arch Vagrant box instead of archlinux/archlinux (no longer available)
    • Install openssl package for Archlinux
    • Removed Ubuntu 20.04 because reached end of life

0.14.0+2.0.2

Note: This a major release update to containerd version 2.0.2! Please read the changelog of containerd v2.0.2 accordingly! In general if you haven't used any "exotic" features so far this version of the Ansible role should cover everything already and upgrading should be smooth. Nevertheless you should test the changes before!

  • POTENTIALLY BREAKING

    • containerd_config variable value was adjusted to to meet containerd configuration requirements of version 3. Please see Full configuration for all possible values. If you haven't adjusted this variable then there should be no need to change anything and upgrading should be smooth.
    • update list of containerd binaries in containerd_binaries variable. containerd-shim-runc-v1 and containerd-shim were removed as no longer provided by upstream.

  • UPDATE

    • update containerd to v2.0.2
    • templates/etc/systemd/system/containerd.service.j2: add dbus.service to After=

  • MOLECULE

    • adjust expected output of ctr pull command

0.13.2+1.7.22

  • UPDATE
    • update containerd to v1.7.22

0.13.1+1.7.20

  • UPDATE
    • update containerd to v1.7.20

0.13.0+1.7.19

  • FEATURE

    • add support for Ubuntu 24.04

  • UPDATE

    • update containerd to v1.7.19

0.12.7+1.7.16

  • UPDATE
    • update containerd to v1.7.16

0.12.6+1.7.15

  • UPDATE

    • update containerd to v1.7.15

  • MOLECULE

    • use alvistack instead of generic Vagrant boxes

0.12.5+1.7.13

  • UPDATE
    • update containerd to v1.7.13

0.12.4+1.7.12

  • add task that creates directory specified in containerd_tmp_directory
  • fix typo in defaults/main.yml
  • Molecule: Change IP addresses
  • Molecule: add two new host variables in molecule/default/molecule.yml

0.12.3+1.7.12

  • update containerd to v1.7.12
  • adjust Github action because of Ansible Galaxy changes

0.12.2+1.7.10

  • update containerd to v1.7.10
  • remove CNI leftovers

0.12.1+1.7.9

  • update containerd to v1.7.9

0.12.0+1.7.8

  • add missing configuration options to make containerd work with Kubernetes out of the box: [plugins."io.containerd.grpc.v1.cri"] -> sandbox_image = "registry.k8s.io/pause:3.8" and [plugins."io.containerd.grpc.v1.cri".containerd.runtimes.runc -> runtime_type = "io.containerd.runc.v2"
  • add "Type": "notify" to containerd_config

0.11.0+1.7.8

Note This release contains quite a few breaking changes. This mostly happened because cri-containerd-*.tar.gz release bundles are deprecated. Also see cri-containerd.DEPRECATED.txt. So containerd doesn't contain CNI plugins and runc anymore. You can still use this bundle if you use version 0.10.0+1.7.3 of this Ansible role. But for this version I decided to remove the support already before containerd v2.0 will be released. That means for containerd_flavor only the value base can be used and k8s has been gone. So that means you have to install CNI - Container Network Interface and runc separately. This should happen before you install containerd. You can use my Ansible roles

Other changes:

  • Breaking containerd_config variable contained a complete dump of all possible containerd configuration settings. The dump was created with containerd config default. So if you just need to execute this command to get all possible configuration options. Since most settings are not relevant I changed the value accordingly and only keep what's different to the default settings. The current settings are intended to be used with a Kubernetes cluster. But in general they should also be a good starting point for standalone installations of containerd. If you had your own containerd_config variable defined this change doesn't affect you.
  • Breaking containerd_flavor variable only supports base value. k8s was removed. This variable most probably will be go away in a later release. If you already used base value here the change wont affect you.
  • Breaking containerd_archive_k8s variable was removed
  • Breaking All runc related resources and variables were removed: containerd_runc_binary_directory, containerd_runc_owner, containerd_runc_group, containerd_runc_binary_mode
  • Breaking All crictl related resources and variables were removed: containerd_crictl_config_file, containerd_crictl_config_directory, containerd_crictl_config_directory_mode, containerd_crictl_config_directory_owner, containerd_crictl_config_directory_group, containerd_crictl_config_file_owner, containerd_crictl_config_file_group, containerd_crictl_config_file_mode, containerd_crictl_config_file_content
  • Breaking All CNI related resources and variables were removed: containerd_cni_binary_directory, containerd_cni_binary_directory_mode, containerd_cni_binary_owner, containerd_cni_binary_group, containerd_cni_binary_mode, containerd_cni_netconfig_file, containerd_cni_netconfig_directory, containerd_cni_netconfig_directory_mode, containerd_cni_netconfig_file_owner, containerd_cni_netconfig_file_group, containerd_cni_netconfig_file_mode, containerd_cni_netconfig_file_content
  • Breaking update to containerd configuration version 2 (that means first line in containerd_config variable is now version = 2)
  • update containerd to v1.7.8
  • Molecule: improve verification step
  • Molecule: increase memory for boxes
  • Molecule: rename kvm scenario to default

0.10.0+1.7.3

  • update containerd to v1.7.3
  • add support for Ubuntu 22.04
  • BREAKING: remove support for Ubuntu 18.04 (reached end of life)
  • BREAKING: When containerd or CNI binaries are upgraded containerd.service will be restarted. Up until this version this only happened when containerd configuration was updated.

0.9.0+1.7.0

  • update containerd to v1.7.0
  • add Molecule verify step

0.8.0+1.6.19

  • update containerd to v1.6.19
  • fix Molecule converge.yml

0.7.0+1.6.14

  • add Github release action to push new release to Ansible Galaxy

0.6.2+1.6.14

  • update containerd to v1.6.14

0.6.1+1.6.9

  • update containerd to v1.6.9

0.6.0+1.6.8

  • fix file permissions in task Downloading containerd archive
  • ansible-lint: fix issues
  • min_ansible_version in meta/main.yml value should be string
  • add .yamllint

0.5.5+1.6.8

  • update containerd to v1.6.8

0.5.4+1.6.6

  • update containerd to v1.6.6

0.5.3+1.6.4

  • update containerd to v1.6.4

0.5.2+1.6.3

  • update containerd to v1.6.3

0.5.1+1.6.2

  • update containerd to v1.6.2

0.5.0+1.6.0

  • update containerd to v1.6.0
  • update containerd_config variable value for containerd v.1.6.0

0.4.0+1.5.9

  • The BinaryName value in container_config variable will now use the value of containerd_runc_binary_directory if defined by default. (contribution by @tiagoblackcode)
  • Ansible's first_found lookup plugin was used to determine the location of modprobe binary. But since all Ansible lookup plugins are only executed on the controller host the modprobe path of the controller host was used in containerd.service file. That's fixed now. (contribution by @tiagoblackcode)

0.3.0+1.5.9

  • In plugins."io.containerd.grpc.v1.cri".containerd.runtimes.runc.options the setting SystemdCgroup is now set to true instead of false. This is relevant for Kubernetes e.g. Also see: Kubernetes container-runtimes - containerd

0.2.1+1.5.9

  • update containerd to v1.5.9

0.2.0+1.5.8

  • add support for Kubernetes by allowing to optionally also install runc, crictl and CNI plugins. containerd builds are available in two flavors: Either just the containerd binaries or containerd binaries plus everything else needed to use containerd together with Kubernetes. Please see defaults/main.yml for all possible settings. With the Kubernetes support a lot of new variables were introduced but for almost all the default values should be good enough. The default containerd_flavor: "base" mimics the behavior of the previous version of this role. So upgrading shouldn't be an issue.

0.1.1+1.5.8

  • update containerd to v1.5.8

0.1.0+1.5.7

  • update containerd to v1.5.7

0.1.0+1.5.5

  • initial commit