Skip to main content
Version: 1.3.x

Istio CNI

By default, Istio injects sidecar proxies into application’s pods in order to handle the traffic for the pod. These sidecars need to be privileged containers as they need to manipulate iptables rules in the pod network namespace to be able to intercept the traffic coming in and out the pod.

This default behavior is not desirable from a security standpoint as it effectively grants the application pods to run with these elevated privileges. The alternative Istio provides to this is the use of a CNI plugin that handles the pod network namespace modifications at pod creation time.

Enable Istio CNI in control plane

In order to enable the Istio CNI plugin in your control plane, you will need to edit the ControlPlane custom resource to include the CNI configuration.

apiVersion: install.tetrate.io/v1alpha1
kind: ControlPlane
metadata:
name: <cluster-name>
namespace: istio-system
spec:
components:
istio:
kubeSpec:
CNI:
chained: true
binaryDirectory: /opt/cni/bin
configurationDirectory: /etc/cni/net.d
traceSamplingRate: 100
hub: <registry-location>
managementPlane:
host: <tsb-address>
port: <tsb-port>
clusterName: <cluster-name>
telemetryStore:
elastic:
host: <elastic-hostname-or-ip>
port: <elastic-port>
version: <elastic-version>

The snippet above shows the default ControlPlane CR with the addition of spec.components.istio.kubeSpec.CNI. This will configure the Istio control plane to deploy the CNI plugin following the provided configuration.

note

Configuration values might change depending on the Kubernetes distribution you use, please refer to the Istio documentation for more information.