Kubernetes Containers

KubernetesClosed Kubernetes, often abbreviated as “K8s”, orchestrates containerized applications to run on a cluster of hosts. is an open-source container orchestration system for automating the deployment, scaling, and management of containerized applications. It operates with a range of container tools and runs containers in a cluster with images built with DockerClosed Docker (specifically, Docker Engine) is a software technology providing operating-system-level virtualization also known as containers., OCIClosed Oracle Cloud Infrastructure - cloud computing platform offered by Oracle Corporation., or Kaniko. It groups containers that make up an application into logical units for easy management and discovery.

Before you can use Kubernetes Containers features in CloudGuard, your Kubernetes cluster must already be onboarded to CloudGuard. See Onboarding Kubernetes Clusters for details on how to do this.

Supported Versions

Name

Version

Kubernetes

  • Version 1.211 and higher (with managed and unmanaged distributions)2

Kubernetes-based Container Orchestration Platforms

  • Red Hat OpenShift v4.6 and higher(Runtime Protection: nodes running Red Hat Enterprise Linux CoreOS)

  • VMware Tanzu TKG v1.2 and higher, TKGI v1.10 and higher

Container Runtime

Node Operating System

Node architecture

  • amd64

1 Kubernetes versions from 1.16 to 1.20 are supported only with HelmClosed A Kubernetes deployment tool for automating creation, packaging, configuration, and deployment of applications and services to Kubernetes clusters. deployment instructions, with the regular helm upgrade --install command.

2 CloudGuard does not support hybrid clusters with multiple (mixed) container runtimes. You cannot change the container runtime after the service is onboarded in the cluster. For this, upgrade the solution.

Notes:

Version Deprecation

Deprecated Kubernetes versions are not supported by cloud vendors and do not get important security updates.

Important - Clusters with deprecated versions can be at risk.

Platform

Service

Supported versions

Microsoft AzureClosed Collection of integrated cloud services that developers and IT professionals use to build, deploy, and manage applications through a global network of data centers managed by Microsoft®.

AKS

Link

Google Cloud Platform

GKE

Link

Amazon Web Services

EKS

Link

Oracle Cloud Infrastructure

OKE

Link

Red Hat

OpenShift

Link

Kubernetes

Kubernetes

Link

Requirements

The CloudGuard agent requires Kubernetes permissions for:

Verb

Group

Resource

Scope

get, list

-

pods

services

nodes

nodes/proxy

serviceaccounts

namespaces

resourcequotas

Cluster-wide

apps

daemonsets

deployments

replicasets

statefulsets

Cluster-wide

networking.k8s.io

networkpolicies

ingresses

Cluster-wide

extensions

ingresses

Cluster-wide

policy

podsecuritypolicies

Cluster-wide

rbac.authorization.k8s.io

roles

rolebindings

clusterroles

clusterrolebindings

Cluster-wide

batch

cronjobs

Cluster-wide

-

pods

secrets

configmaps

Agent namespace (default: checkpoint)

patch

admissionregistration.k8s.io

validatingwebhookconfigurations

Cluster-wide

all (*)

*.cloudguard.checkpoint.com

all (*)

Agent namespace (default: checkpoint)

The CloudGuard agent requires these Kubernetes permissions for OpenShift Kubernetes clusters:

Verb

Group

Resource

Scope

get, list

config.openshift.io

clusteroperators (resourceName: openshift-apiserver)

Cluster-wide

operator.openshift.io

openshiftapiservers, kuberapiservers (resourceName: cluster)

Cluster-wide

security.openshift.io

securitycontextconstraints

Cluster-wide

get

-

-

configmaps (resourceName: config)

openshift-kube-controller-manager, openshift-apiserver, openshift-kube-apiserver

configmaps (resourceName: kube-scheduler-podClosed The smallest and simplest Kubernetes object. A pod represents a set of running containers on your cluster. A pod is typically set up to run a single primary container. It can also run optional sidecar containers that add supplementary features like logging. Pods are commonly managed by a Deployment.)

openshift-kube-scheduler

Note - For Kubernetes clusters on OpenShift, CloudGuard creates additional roles in OpenShift namespaces. The role bindings bind these roles to the CloudGuard service account (in CloudGuard namespace used for the agent).

Linux allows assigning specific capabilities to processes, thus restricting the processes to only the minimum privileges required to perform their tasks and reducing the risk of security breaches.

The CloudGuard agents require these Linux kernel capabilities:

Agent Name

Container Capabilities

Host Network

Privileged
Runtime Protection (runtime-daemon)

  • SYS_RESOURCE

  • SYS_ADMIN

  • SYS_NICE

  • SYS_PTRACE

  • FOWNER

  • SYS_PACCT

  • NET_ADMIN

  • NET_RAW

  • AUDIT_READ

  • AUDIT_WRITE

  • AUDIT_CONTROL

 required

OpenShift, CRI-O, CoreOS, Bottlerocket:

required
Flow Logs(flowlogs-daemon)

  • SYS_RESOURCE

  • SYS_ADMIN

  • NET_ADMIN

 required

OpenShift: required
Image Assurance (imagescan-daemon)

  • NET_BIND_SERVICE

 not required

OpenShift and CRI-O:

required

CloudGuard agents must have connectivity to these domains:

Blade or Agent Address

CloudGuard

Image Assurance

Image Scan

.dome9.com

Notes:

  • The domain .dome9.com represents Check Point's Network Domain Objects.

  • For Image Assurance agents ver. 2.28.0 and lower, add the .checkpoint.com domain.

Runtime Protection

https://storage.googleapis.com/cos-tools

https://rep.checkpoint.com/file-rep/service/v2.0/query

Container RegistryClosed A collection of repositories used to store and access container images.

https://quay.io/checkpoint

Instead of the domain objects, you can use the region-specific URLs for your Data Center location from the table below. Add these endpoints to the allowlist.

Blade or Agent Address

Runtime Protection

https://storage.googleapis.com/cos-tools

https://rep.checkpoint.com/file-rep/service/v2.0/query

Container Registry

https://quay.io/checkpoint

For Image Scan agents ver 2.28.0 and lower, you must use additional endpoints. To learn more about agent's version, see Agent Version Life Cycle.

Blade or Agent Address

Image Assurance

https://rpm-serv.sg.iaas.checkpoint.com

https://shiftleft.portal.checkpoint.com/

Image Scan

https://shiftleft-prod-bucket.sg.iaas.checkpoint.com
Blade or Agent Address

CloudGuard

https://api-cpx.dome9.com

https://api.dome9.com

Threat Intelligence

https://validator-prod-k8s.s3.amazonaws.com

Image Scan

(agent ver. 2.28.0 and lower)

https://us-gw.sg.iaas.checkpoint.com
Blade or Agent Address

CloudGuard

https://api-cpx.eu1.dome9.com

https://api.eu1.dome9.com

Threat Intelligence

https://validator-prod-533924475734-k8s.s3.eu-west-1.amazonaws.com

Image Scan

(agent ver. 2.28.0 and lower)

https://eu-gw.sg.iaas.checkpoint.com
Blade or Agent Address

CloudGuard

https://api-cpx.ap2.dome9.com

https://api.ap2.dome9.com

Threat Intelligence

https://validator-prod-583664506098-k8s.s3.ap-southeast-2.amazonaws.com

Image Scan

(agent ver. 2.28.0 and lower)

https://au-gw.sg.iaas.checkpoint.com
Blade or Agent Address

CloudGuard

https://api-cpx.cace1.dome9.com

https://api.cace1.dome9.com

Threat Intelligence

https://validator-prod-052001227150-k8s.ca-central-1.amazonaws.com

Image Scan

(agent ver. 2.28.0 and lower)

https://ca-gw.sg.iaas.checkpoint.com
Blade or Agent Address

CloudGuard

https://api-cpx.ap3.dome9.com

https://api.ap3.dome9.com

Threat Intelligence

https://validator-prod-573281234161-k8s.s3.ap-south-1.amazonaws.com

Image Scan

(agent ver. 2.28.0 and lower)

https://in-gw.sg.iaas.checkpoint.com

If the CloudGuard pod image is uploaded to a private repository, connectivity to Container Registry is not necessary. In this case, the Helm chart parameter image.repository must be changed to indicate the location of the image. For more information about how to set this parameter, see https://github.com/CheckPointSW/charts/tree/master/checkpoint/cloudguard.

Each CloudGuard feature can have pods that run in daemonsets and pods that run in deployments. For the pods in daemonsets, resources are shown in the table below per node. For the pods in deployments, resources are below per cluster. The pods that run in deployments can certainly run on different nodes.

You can find the default values of requests and limits in the defaults.yaml on the Helm Chart.

Basic Feature

Per cluster or node

CPU (millicores)

Memory (MiB)

requests

limits

requests

limits

Posture Management

per cluster

100

200

50

50

Image Assurance

per cluster

200

1050

200

2600

per node - Docker

50

50

50

50

per node - containerd

200

250

150

150

per node - CRI-O*

200

300

250

250

Admission Control

per cluster

1150

1350

330

450

* OpenShift uses the CRI-O runtime

Note - Image Assurance scanning engine (imagescan-engine) requires free ephemeral storage, which size is double maximal size of the scanned image. Maximal image size means the uncompressed, docker save output image size.

Premium Feature

Per cluster or node

CPU (millicores)

Memory (MiB)

requests

limits

requests

limits

Runtime Protection

per cluster

50

50

30

50

per node*

200

400

300

800

Flow Logs

per node

100

200

30

100

* Large Nodes (above 8 vCPUs) may require additional resources in case of pod restarts.

More Links

For Kubernetes terminology, see the Glossary in the Kubernetes documentation.