Add container checkpoint support

Signed-off-by: tu1h <[email protected]>

docs/vars.md

@@ -272,6 +272,8 @@ node_taints:
   * `audit_webhook_batch_max_size`: 100
   * `audit_webhook_batch_max_wait`: 1s
 
+* *container_checkpoint_enabled* - When set to `true`, enables the Checkpoint API on kubelet and container runtime(only cri-o yet). It's required to install [CRIU](https://criu.org/Installation) on the host when dumping checkpoints. You can follow the [documentation](https://kubernetes.io/blog/2022/12/05/forensic-container-checkpointing-alpha/). **Note**: It's still  in experimental stage, just for container analytics so far. 
+
 ### Custom flags for Kube Components
 
 For all kube components, custom flags can be passed in. This allows for edge cases where users need changes to the default deployment that may not be applicable to all deployments.

roles/container-engine/cri-o/tasks/main.yaml

@@ -175,6 +175,13 @@
   notify: Restart crio
   when: http_proxy is defined or https_proxy is defined
 
+- name: Cri-o | write cri-o options drop-in
+  template:
+    src: options.conf.j2
+    dest: /etc/systemd/system/crio.service.d/options.conf
+    mode: 0644
+  notify: Restart crio
+
 - name: Cri-o | configure the uid/gid space for user namespaces
   lineinfile:
     path: '{{ item.path }}'

roles/container-engine/cri-o/templates/options.conf.j2

@@ -0,0 +1,2 @@
+[Service]
+Environment=CRIO_CONFIG_OPTIONS={% if container_checkpoint_enabled is defined and container_checkpoint_enabled %}--enable-criu-support=true{% endif %}

roles/kubernetes/control-plane/templates/kubeadm-config.v1beta3.yaml.j2

@@ -447,10 +447,13 @@ clusterDNS:
 {% for dns_address in kubelet_cluster_dns %}
 - {{ dns_address }}
 {% endfor %}
+{% if container_checkpoint_enabled is defined and container_checkpoint_enabled -%}
+{% set kubelet_feature_gates = kubelet_feature_gates + ["ContainerCheckpoint=true"] -%}
+{% endif -%}
 {% if kubelet_feature_gates or kube_feature_gates %}
 {% set feature_gates = ( kubelet_feature_gates | default(kube_feature_gates, true) ) %}
 featureGates:
-{%   for feature in feature_gates %}
+{%   for feature in (feature_gates | unique) %}
   {{ feature | replace("=", ": ") }}
 {%   endfor %}
 {% endif %}

roles/kubernetes/node/templates/kubelet-config.v1beta1.yaml.j2

@@ -120,9 +120,12 @@ resolvConf: "{{ kube_resolv_conf }}"
 {% if inventory_hostname in groups['kube_node'] and kubelet_node_config_extra_args %}
 {{ kubelet_node_config_extra_args | to_nice_yaml(indent=2) }}
 {% endif %}
+{% if container_checkpoint_enabled is defined and container_checkpoint_enabled -%}
+{% set kubelet_feature_gates = kubelet_feature_gates + ["ContainerCheckpoint=true"] -%}
+{% endif -%}
 {% if kubelet_feature_gates or kube_feature_gates %}
 featureGates:
-{% for feature in (kubelet_feature_gates | default(kube_feature_gates, true)) %}
+{% for feature in (kubelet_feature_gates | default(kube_feature_gates, true) | unique) %}
   {{ feature | replace("=", ": ") }}
 {% endfor %}
 {% endif %}

roles/kubespray-defaults/defaults/main.yaml

@@ -675,3 +675,6 @@ sysctl_file_path: "/etc/sysctl.d/99-sysctl.conf"
 
 system_upgrade: false
 system_upgrade_reboot: on-upgrade  # never, always
+
+# Forensic container checkpointing in Kubernetes. It's experimental so far. See [Checkpoint API](https://kubernetes.io/docs/reference/node/kubelet-checkpoint-api/)
+container_checkpoint_enabled: false