Verify mutual TLS is enabled

Verify that mTLS is active between workloads in ambient mesh using ztunnel config, traffic inspection, or ztunnel logs.

With ambient mesh, communication between your workloads is automatically secured via mTLS as soon as you add them to the mesh. No additional configuration is required.

To validate that traffic is secured via mTLS, choose from multiple methods.

Confirm via ztunnel config

Using the istioctl ztunnel-config workloads command, you can quickly view whether your workload is configured to send and accept HBONE traffic.

In the following example, all pods except for the kubernetes pod are configured with the HBONE protocol.

istioctl ztunnel-config workloads

NAMESPACE    POD NAME                                IP         NODE                     WAYPOINT PROTOCOL
default      details-v1-857849f66-ft8wx              10.42.0.5  k3d-k3s-default-agent-0  None     HBONE
default      kubernetes                              172.20.0.3                          None     TCP
default      productpage-v1-c5b7f7dbc-hlhpd          10.42.0.8  k3d-k3s-default-agent-0  None     HBONE
default      ratings-v1-68d5f5486b-b5sbj             10.42.0.6  k3d-k3s-default-agent-0  None     HBONE
default      reviews-v1-7dc5fc4b46-ndrq9             10.42.1.5  k3d-k3s-default-agent-1  None     HBONE
default      reviews-v2-6cf45d556b-4k4md             10.42.0.7  k3d-k3s-default-agent-0  None     HBONE
default      reviews-v3-86cb7d97f8-zxzl4             10.42.1.6  k3d-k3s-default-agent-1  None     HBONE

Confirm with metrics

If you installed Prometheus, open the Prometheus UI:

istioctl dashboard prometheus

In Prometheus, you can view TCP metrics for your traffic. First, select Graph. Then, enter any of the following metrics: istio_tcp_connections_opened_total, istio_tcp_connections_closed_total, istio_tcp_received_bytes_total, or istio_tcp_sent_bytes_total. Lastly, click Execute. The data contains entries such as:

istio_tcp_connections_opened_total{
  app="ztunnel",
  connection_security_policy="mutual_tls",
  destination_principal="spiffe://cluster.local/ns/default/sa/bookinfo-details",
  destination_service="details.default.svc.cluster.local",
  reporter="source",
  request_protocol="tcp",
  response_flags="-",
  source_app="curl",
  source_principal="spiffe://cluster.local/ns/default/sa/curl",source_workload_namespace="default",
  ...}

Validate that the connection_security_policy value is set to mutual_tls along with the expected source and destination identity information.

Confirm with ztunnel logs

Check the source or destination ztunnel logs to confirm that mTLS is enabled between your workloads and that peer identities for the source and destination exist. Below is an example of the source ztunnel logs for a request from the curl service to the details service:

2024-08-21T15:32:05.754291Z info access connection complete src.addr=10.42.0.9:33772 src.workload="curl-7656cf8794-6lsm4" src.namespace="default"
src.identity="spiffe://cluster.local/ns/default/sa/curl" dst.addr=10.42.0.5:15008 dst.hbone_addr=10.42.0.5:9080 dst.service="details.default.svc.cluster.local"
dst.workload="details-v1-857849f66-ft8wx" dst.namespace="default" dst.identity="spiffe://cluster.local/ns/default/sa/bookinfo-details"
direction="outbound" bytes_sent=84 bytes_recv=358 duration="15ms"

Validate the src.identity and dst.identity values are correct. These are the identities used for the mTLS communication between the source and destination workloads.

Confirm with Kiali

If you have Kiali and Prometheus installed, you can visualize your workload communication in the ambient mesh using Kiali’s dashboard. The padlock icon on a connection between workloads indicates that mTLS is enabled, along with the peer identity information:

Confirm via tcpdump

If you have access to your Kubernetes worker nodes, run the tcpdump command to capture all traffic on the network interface, with optional focus on the application and HBONE ports. In this example, 9080 is the port of the details service and 15008 is the HBONE port:

tcpdump -nAi eth0 port 9080 or port 15008

If you do not have access to the worker nodes, use the netshoot container image to run tcpdump:

POD=$(kubectl get pods -l app=details -o jsonpath="{.items[0].metadata.name}")
kubectl debug $POD -i --image=nicolaka/netshoot -- tcpdump -nAi eth0 port 9080 or port 15008