Linux on Kubernetes

The Sysdig Helm chart sysdig-deploy includes configuration options for customizing the agent’s behavior and integrating with other Sysdig components. By using the Helm chart, you can easily deploy the Sysdig Agent on Kubernetes and take advantage of Sysdig’s powerful monitoring and security capabilities.

If you are migrating from previously installed Sysdig components to Cluster Shield, see Cluster Shield.

Prerequisites

• Review the Installation Requirements before getting started.

Component

• Kubectl installed
• Helm v3.8 and above
• Your agent access key
• Sysdig SaaS region

Use the Quick Start Wizard

1. Log in to Sysdig Secure as an administrator and select Integrations > Data Sources|Sysdig Agent.

2. Click +Add Agent and select Kubernetes. The Helm installation method is recommended.

3. As prompted by the Wizard screen, enter:

   • Cluster Name: Specify the name of your Kubernetes cluster.

4. The Wizard will auto-populate a code snippet from the cluster name, along with autodetected Sysdig Secure endpoint and agent access key information.

5. Copy and run the Helm commands. This will install the Cluster Shield and give you:

   • Runtime Threat Detection
   • Vulnerability Host Scanning
   • Runtime Image Scanning
   • Kubernetes Security Posture Management (Compliance)

Otherwise, you can use the generic Helm commands detailed below and customize as needed.

Use Command Line

Full Deploy

This command replicates the Wizard process and installs the Sysdig Cluster Shield and the additional components to deliver runtime threat detection, host scanning, runtime image scanning, and compliance, or Kubernetes Security Posture Management (KSPM).

It uses the Helm chart sysdig-deploy.

• Helm
• values.sysdig.yaml

helm repo add sysdig https://charts.sysdig.com
helm repo update
helm install -n sysdig-agent --create-namespace \
  --set global.clusterConfig.name=<CLUSTER_NAME> \
  --set global.sysdig.region=<SAAS_REGION> \
  --set global.sysdig.accessKey=<ACCESS_KEY> \
  --set global.kspm.deploy=true \
  --set nodeAnalyzer.enabled=true \
  --set nodeAnalyzer.nodeAnalyzer.benchmarkRunner.deploy=false \
  --set nodeAnalyzer.nodeAnalyzer.hostAnalyzer.deploy=false \
  --set nodeAnalyzer.nodeAnalyzer.runtimeScanner.deploy=false \
  --set nodeAnalyzer.nodeAnalyzer.imageAnalyzer.deploy=false \
  --set nodeAnalyzer.nodeAnalyzer.hostScanner.deploy=true \
  --set kspmCollector.enabled=false \
  --set admissionController.enabled=false \
  --set clusterShield.enabled=true \
  --set clusterShield.cluster_shield.log_level=warn \
  --set clusterShield.cluster_shield.features.audit.enabled=true \
  --set clusterShield.cluster_shield.features.container_vulnerability_management.enabled=true \
  --set clusterShield.cluster_shield.features.posture.enabled=true \
sysdig sysdig/sysdig-deploy

  

## Create a values.sysdig.yaml file with the following:
global:
  clusterConfig:
    name: <CLUSTER_NAME>
  sysdig:
    region: <SAAS_REGION>
    accessKey: <ACCESS_KEY>
  kspm:
    deploy: true
nodeAnalyzer:
  enabled: true
  nodeAnalyzer:
    benchmarkRunner:
      deploy: false
    hostAnalyzer:
      deploy: false
    runtimeScanner:
      deploy: false
    imageAnalyzer:
      deploy: false
    hostScanner:
      deploy: true
kspmCollector:
  enabled: false
admissionController:
  enabled: false
clusterShield:
  enabled: true
  cluster_shield:
    log_level: warn
    features:
      audit:
        enabled: true
      container_vulnerability_management:
        enabled: true
      posture:
        enabled: true

## Deploy the Chart

  helm repo add sysdig https://charts.sysdig.com
  helm install -n sysdig-agent sysdig sysdig/sysdig-deploy -f values.sysdig.yaml

 

Runtime Threat Detection Only

To install only Runtime Threat Detection (Sysdig Agent), use:

• Helm
• values.sysdig.yaml

helm repo add sysdig https://charts.sysdig.com
helm install sysdig-agent --namespace sysdig-agent --create-namespace \
    --set global.sysdig.accessKey=<ACCESS_KEY> \
    --set global.sysdig.region=<SAAS_REGION> \
    --set nodeAnalyzer.enabled=false \
    --set global.clusterConfig.name=<CLUSTER_NAME> \
    sysdig/sysdig-deploy
    
  

## Create a values.sysdig.yaml file with the Following:
global:
  sysdig:
    accessKey: <ACCESS_KEY>
    region: <SAAS_REGION>
  clusterConfig:
    name: <CLUSTER_NAME>
nodeAnalyzer:
  enabled: false

## Then, install with the following:

helm repo add sysdig https://charts.sysdig.com
helm install -n sysdig-agent sysdig sysdig/sysdig-deploy -f values.sysdig.yaml

 

Pod Security Admission

If you’re enforcing PSA, add the privileged policy to the sysdig-agent namespace:

kubectl label --overwrite ns sysdig-agent pod-security.kubernetes.io/enforce=privileged

Parameter Definitions

The command above specifies several options:

• --namespace sysdig-agent

  • Specifies that the Sysdig deployment should be installed in the “sysdig-agent” namespace.

• --set global.sysdig.accessKey=<ACCESS_KEY>

  • Specifies the Sysdig access key to use when connecting to the Sysdig backend. Replace <ACCESS_KEY> with your actual access key.

• --set global.sysdig.region=<SAAS_REGION>

  • Specifies the Sysdig region to use. Replace <SAAS_REGION> with the region where your Sysdig account is located.

• --set nodeAnalyzer.secure.vulnerabilityManagement.newEngineOnly=true

  • Enables the new engine for vulnerability management in Sysdig Secure.

• --set global.kspm.deploy=true

  • Enables the deployment of the KSPM Collector and Analyzer components.

• --set nodeAnalyzer.nodeAnalyzer.benchmarkRunner.deploy=false

  • Disables the deployment of the legacy Node Analyzer Benchmark Runner component.

• --set nodeAnalyzer.nodeAnalyzer.hostScanner.deploy=true

  • Installs the Host Scanner.

• --set global.clusterConfig.name=<CLUSTER_NAME>

  • Specifies the name of your Kubernetes cluster. Replace <CLUSTER_NAME> with your actual cluster name.

• --set global.sysdig.tags.<myKey>=<myValue>

  • Optionally, add user-defined tag by defining a Key:Value pair. Replace <myKey> with your tag key and <myValue> with your tag value. you could add multiple tags by adding another row per tag.

After running these commands, the Sysdig Cluster Shield and associated components should be installed and running on your Kubernetes cluster, and will begin sending data to the Sysdig backend for analysis and monitoring.

About Host Scanner

The runtime scanner and host scanner are deployed by default with the given Helm commands.

Opting Out

If for some reason you don’t want to use host scanning, you can opt-out using the Helm chart flag:

--set nodeAnalyzer.nodeAnalyzer.hostScanner.deploy=false

Specific Kubernetes Platforms

OpenShift, GKE, OKE, and MKE

If you are using OpenShift, Google Kubernetes Engine (GKE) Standard, Oracle Kubernetes Engine (OKE), or Mirantis Kubernetes engine (MKE), enable eBPF with the following option:

• --set agent.ebpf.enabled=true

GKE Autopilot

If you are using GKE autopilot, enable the following option:

• -–set agent.gke.autopilot=true

Rancher Kubernetes Engine 1

Rancher Kubernetes Engine V1 (RKE1) internally changes container names at the Docker layer in Kubernetes. For instance, while a container may be named nginx in your cluster, its underlying name in the Docker engine might resemble k8s_nginx_nginx_test-container-id-b277-3066f0a5-7115-48f0-bb2d-c48f71663087_9d80f7b5-1827-4a05-83. The Sysdig Agent communicates with the Docker engine to retrieve container names, and the internal names will be displayed in the Sysdig UI instead of the container name provided within the cluster.

Rancher Kubernetes Engine 2

Rancher Kubernetes Engine V2(RKE2) uses a different containerd socket path, /run/k3s/containerd/containerd.sock, instead of the expected /var/run/containerd/containerd.sock. To accommodate this, ensure that you mount the alternative path into the runtime scanner via the chart’s values. This is necessary for the scanner to connect to the containerd socket and identify running containers for subsequent image scanning. Update the chart’s values accordingly.

Using the Key-Value Pair

--set nodeAnalyzer.nodeAnalyzer.extraVolumes.volumes[0].name=socketpath \
--set nodeAnalyzer.nodeAnalyzer.extraVolumes.volumes[0].hostPath.path=/run/k3s/containerd/containerd.sock \
--set nodeAnalyzer.nodeAnalyzer.runtimeScanner.extraMounts[0].name=socketpath \
--set nodeAnalyzer.nodeAnalyzer.runtimeScanner.extraMounts[0].mountPath=/var/run/containerd/containerd.sock \

Using values.sysdig.yaml

nodeAnalyzer:
  nodeAnalyzer:
    extraVolumes:
      volumes:
        - name: socketpath
          hostPath:
            path: /run/k3s/containerd/containerd.sock
    runtimeScanner:
      extraMounts:
        - name: socketpath
          mountPath: /var/run/containerd/containerd.sock

Mirantis Kubernetes Engine

The Mirantis Kubernetes Engine uses Docker as its runtime, so to accommodate this, make sure you mount the Docker application path in the Agent via the chart’s values. This is necessary in order for the Agent to have malware detection and malware prevention to work properly. Update the chart’s values accordingly.

Using the Key-Value Pair

--set agent.extraVolumes.volumes[0].name=sysdig-docker-apps \
--set agent.extraVolumes.volumes[0].hostPath.path=/opt/apps/docker \
--set agent.extraVolumes.mounts[0].mountPath=/host/opt/apps/docker \
--set agent.extraVolumes.mounts[0].name=sysdig-docker-apps \
--set agent.extraVolumes.mounts[0].readOnly=true

Using values.sysdig.yaml

agent:
  extraVolumes:
    volumes:
      - name: sysdig-docker-apps
        hostPath:
          path: /opt/apps/docker
    mounts:
      - mountPath: /host/opt/apps/docker
        name: sysdig-docker-apps
        readOnly: true

Additional Volumes

Additional volumes might be required in the Sysdig Agent for mapping ConfigMaps, Secrets or additional host paths. In some specific cases, such as SLES15 on Rancher, the proper ld-linux-* library is under the host /lib64 so the kernel module build fails. To handle, add a specific volume mount /lib64 to /host/usr/lib64. For example:

agent:
  daemonset:
    kmodule:
      extraVolumes:
        volumes:
          - name: lib64-vol
            hostPath:
              path: /lib64
        mounts:
          - mountPath: /host/usr/lib64
            name: lib64-vol

While for the main Agent container you can specify the extraVolumes in the agent root like:

agent:
  extraVolumes:
    volumes:
      - name: repo-new-cm
        configMap:
          name: my-cm
          optional: true
      - name: repo-new-secret
          secret:
          secretName: my-secret
    mounts:
      - mountPath: mount-path
        name: repo-new-cm
      - mountPath: mount-path
        name: repo-new-secret

For additional configuration options, including on-premises or using a proxy, see the sysdig-deploy readme.

You can also use Helm to

Install Kubernetes Audit Logging (Admission Controller)

Install Rapid Response

Deploy Sysdig Agent on Large Kubernetes Clusters (Tech Preview)

Sysdig Agent generates a significant load when running on large Kubernetes clusters to fetch the Kubernetes metadata used to enrich metrics and events. You can reduce this load with Backend Enrichment, whereby only Cluster Shield (or the delegated agent) collects the Kubernetes metadata and the data enrichment happens in the backend. We recommend you enable Backend Enrichment particularly for large clusters (>500 nodes), but it can be beneficial for smaller clusters too.

Enable Backend Enrichment

Backend enrichment enablement in SaaS environments is controlled by Sysdig. Contact Sysdig Support to enable it in your environment.

Prerequisites

• Linux Agent version 13.0.0 or later.

• The Agent configuration be_enrichment.enabled set to true (default) to enable the negotiation with the Sysdig Backend.

• K8S_NODE environment variable set to the Kubernetes node name on the Sysdig Agent container. This is the default behavior when using the Sysdig Helm charts.

• Sysdig strongly recommends deploying cluster-shield and enabling the kubernetes_metadata feature for sending the metadata used for backend enrichment.

  This approach is preferred over using delegated agents.

Deploy with Helm Chart

helm install sysdig-agent \
  --namespace sysdig-agent \
  --set agent.sysdig.settings.be_enrichment=true \
  --set agent.sysdig.settings.k8s_delegated_nodes=0 \
  --set clusterShield.enabled=true \
  --set clusterShield.cluster_shield.features.kubernetes_metadata.enabled: true \
  sysdig/sysdig-deploy

Additional Agent Configuration

For additional configuration settings, refer to the following list:

be_enrichment:
  enabled: true # Default since Agent version 13.0.0.
  override: true # (Optional) Enables backend enrichment mode in the Agent side regardless of the backend enablement.
  enforce: true # (Optional) The agent fails during the startup if backend enrichment is not enabled on the Sysdig backend side.
k8s_delegated_nodes: 0 # Set the number of delegated agents to 0 when KubernetesMetadata component is used
k8s_add_labels_to_container: true # (Optional) Add the extracted Kubernetes labels to containers during runtime. Useful when Local Forwarder or Audit tap are used. Setting the number of delegated nodes to 0 assumes that Cluster Shield is deployed with the Kubernetes Metadata component enabled.

Advantages

• Offers improved scalability and performance in large Kuberentes clusters, minimizing the load the Agents put on the Kubernetes API server.

• Enhances the agent’s efficiency, particularly by reducing memory usage for caching Kubernetes metadata and minimizing egress traffic due to smaller metrics messages. The performance gain is proportional to the size and density of the cluster.

• Helps having a more uniform host agent resource consumption by eliminating the need for delegated agents.

Known Limitations

• NetSec is unsupported when Backend Enrichment is enabled. This feature relies on the local availability of Kubernetes metadata in each host agent, which might not function as expected in this configuration.
• A reduced set of Kubernetes labels is available when Local Forwarder or Audit Tap are used.
• The configuration for event custom labels is currently not supported.