README.md

ziti-controller

Host an OpenZiti controller in Kubernetes

Requirements

Repository	Name	Version
https://charts.jetstack.io	cert-manager	~1.11.0
https://charts.jetstack.io	trust-manager	~0.7.0
https://kubernetes.github.io/ingress-nginx/	ingress-nginx	~4.5.2

Overview

This chart runs a Ziti controller in Kubernetes. It uses the custom resources provided by cert-manager and trust-manager, i.e., Issuer, Certificate, and Bundle. Delete the controller pod after an upgrade for the new controller configuration to take effect.

Here's a checklist that must be satisfied for each of the controller's TLS servers: router control plane (ctrlPlane), each web listener(s) (clientApi, at least).

1. Each of the controller's TLS servers must have an advertised host (FQDN) and TCP port. The chart ensures the DNS SAN is added for this address. The port is always 443 if ingress is enabled for ClusterIP service.
2. A K8s Service must be configured and reachable by all clients. (i.e., type ClusterIP+Ingress, NodePort, or LoadBalancer)
3. Each K8s Service must be configured for TLS passthrough (i.e., server TLS must not be terminated by an intermediary LB).
4. All clients must access the controller through the advertised address(es).

Requirements

Add the OpenZiti Charts Repo to Helm

helm repo add openziti https://docs.openziti.io/helm-charts/

This chart requires Certificate, Issuer, and Bundle resources to be applied before installing the chart. Sub-charts cert-manager, and trust-manager will be installed automatically. You may disable the sub-charts if you wish to provide these resources separately, but if you do so then please use the sub-chart values at the foot of Values.yaml to ensure those charts are correctly configured.

Install Required Custom Resource Definitions

This step satisfies Helm's requirement that the CRDs used in the umbrella chart already exist in Kubernetes before installing the controller chart.

kubectl apply -f https://github.com/cert-manager/cert-manager/releases/latest/download/cert-manager.crds.yaml
kubectl apply -f https://raw.githubusercontent.com/cert-manager/trust-manager/v0.9.0/deploy/crds/trust.cert-manager.io_bundles.yaml

Minimal Installation

This first example shows a minimal installation for a Kubernetes distribution that provides TLS pass-through for Service type LoadBalancer, e.g., k3s, k3d, Minikube. This is useful for environments where there's no cost, or justifiable cost, associated with provisioning a LoadBalancer with TLS passthrough.

You must supply one value when you install the chart.

Key	Type	Default	Description
clientApi.advertisedHost	string	nil	the DNS name that edge clients and routers will resolve to reach this controller's edge client API
clientApi.advertisedPort	string	nil	the TCP port associated with the advertisedHost to advertise to edge clients and routers

helm install \
    --namespace ziti-controller ziti-controller-minimal1 \
    openziti/ziti-controller \
        --set clientApi.advertisedHost="ziti-controller-minimal.example.com" \
        --set clientApi.advertisedPort="443"

A default admin user and password will be generated and saved to a secret during installation. The credentials can be retrieved using this command:

kubectl get secret \
    -n ziti-controller ziti-controller-minimal1-admin-secret \
    -o go-template='{{range $k,$v := .data}}{{printf "%s: " $k}}{{if not $v}}{{$v}}{{else}}{{$v | base64decode}}{{end}}{{"\n"}}{{end}}'

You may log in the ziti CLI with one command or omit the -p part to prompt:

ziti edge login ziti-controller-minimal.example.com:1280 \
    --yes \
    --username admin \
    --password $(
        kubectl -n ziti-controller \
            get secrets ziti-controller-minimal1-admin-secret \
                -o go-template='{{index .data "admin-password" | base64decode }}'
        )

Using ClusterIP Services with an Ingress Controller

The default K8s service type for this chart is ClusterIP. You can publish these cluster-internal services with an Ingress resource. You need an Ingress Controller. Here's an example of using the community ingress-nginx chart to provision ingresses for the controller's ClusterIP services.

Ensure you have the ingress-nginx chart installed with controller.extraArgs.enable-ssl-passthrough=true. You can verify this feature is enabled by running kubectl describe pods {ingress-nginx-controller pod} and checking the args for --enable-ssl-passthrough=true.

If necessary, patch the ingress-nginx deployment to enable TLS passthrough.

kubectl patch deployment "ingress-nginx-controller" \
    --namespace ingress-nginx \
    --type json \
    --patch '[{"op": "add",
        "path": "/spec/template/spec/containers/0/args/-",
        "value":"--enable-ssl-passthrough"
    }]'

Create a Helm chart values file like this.

# /tmp/controller-values.yml
clientApi:
    advertisedHost: ziti-controller-managed.example.com
    advertisedPort: 443
    service:
        type: ClusterIP
    ingress:
        enabled: true
        ingressClassName: nginx
        annotations:
            kubernetes.io/ingress.allow-http: "false"
            nginx.ingress.kubernetes.io/ssl-passthrough: "true"

Now install or upgrade this controller chart with your values file.

helm install \
    --namespace ziti-controller ziti-controller-managed1 \
    openziti/ziti-controller \
    --values /tmp/controller-values.yml

Expose the Router Control Plane

This is applicable if you have any routers outside the Ziti controller's cluster. You must configure pass-through TLS LoadBalancer or Ingress for the control plane service. Routers running in the same cluster as the controller can use the cluster service named {controller release}-ctrl (the "ctrl" endpoint). This example demonstrates a pass-through Ingress resource for nginx-ingress.

Merge this with your Helm chart values file before installing or upgrading.

ctrlPlane:
    advertisedHost: ziti-controller-managed-ctrl.example.com
    advertisedPort: 443
    service:
        enabled: true
    ingress:
        enabled: true
        ingressClassName: nginx
        annotations:
            kubernetes.io/ingress.allow-http: "false"
            nginx.ingress.kubernetes.io/ssl-passthrough: "true"

Extra Security for the Management API

You can split the client and management APIs into separate cluster services by setting managementApi.service.enabled=true. With this configuration, you'll have an additional cluster service named {controller release}-mgmt that is the management API, and the client API will not have management features.

This Helm chart's values allow for both operational scenarios: combined and split. The default choice is to expose the combined client and management APIs as the cluster service named {controller release}-client, which is convenient because you can use the ziti CLI immediately. For additional security, you may shelter the management API by splitting these two sets of features, exposing them as separate API servers. After the split, you can access the management API in several ways:

• running the web console inside the cluster,
• hosting a Ziti service, or
• kubectl port-forward.

Advanced PKI

The default configuration generates a singular PKI root of trust for all the controller's servers and the edge signer CA. Optionally, you may provide the name of a cert-manager Issuer or ClusterIssuer to become the root of trust for the Ziti controller's identity.

Merge this with your Helm chart values file before installing or upgrading.

ctrlPlane:
    issuer:
        kind: ClusterIssuer
        name: my-alternative-cluster-issuer

You may also configure the Ziti controller to use separate PKI roots of trust for its three main identities: control plane, edge signer, and web bindings.

For example, to use a separate CA for the edge signer function, merge this with your Helm chart values file before installing or upgrading.

edgeSignerPki:
    enabled: true

Prometheus Monitoring

This chart provides a default disabled ziti-controller-prometheus k8s service for prometheus, which can be enabled with prometheus.service.enabled. Enabling it will create a prometheus ServiceMonitor for configuring the prometheus endpoint. It is also important that you enable fabric.events.enabled for getting a full set of metrics.

For more information, please check here.

Values Reference

Key	Type	Default	Description
additionalConfigs	object	{}	allow the insertion of raw/unmodified config blocks Additional configs will be appended to the top-level keys of the ziti controller config ( events, edge, web, network, ctrl) These configs allow for OVERRIDING behavior in any top level config block
additionalVolumes	list	[]	additional volumes to mount to ziti-controller container
affinity	object	{}	deployment template spec affinity
ca.clusterDomain	string	"cluster.local"	Set a custom cluster domain if other than cluster.local
ca.duration	string	"87840h"	Go time.Duration string format
ca.renewBefore	string	"720h"	Go time.Duration string format
cert-manager.enableCertificateOwnerRef	bool	true	clean up secret when certificate is deleted
cert-manager.enabled	bool	false	install the cert-manager subchart
cert-manager.installCRDs	bool	false	CRDs must be applied in advance of installing the parent chart
cert.duration	string	"87840h"	server certificate duration as Go time.Duration string format
cert.renewBefore	string	"720h"	rewnew server certificates before expiry as Go time.Duration string format
clientApi.advertisedHost	string	nil	global DNS name by which routers can resolve a reachable IP for this service
clientApi.advertisedPort	int	443	cluster service, node port, load balancer, and ingress port
clientApi.containerPort	int	1280	cluster service target port on the container
clientApi.dnsNames	list	[]	additional DNS SANs
clientApi.ingress.annotations	string	nil	ingress annotations, e.g., to configure ingress-nginx
clientApi.ingress.enabled	bool	false	create an ingress for the cluster service
clientApi.service.enabled	bool	true	create a cluster service for the deployment
clientApi.service.type	string	"LoadBalancer"	expose the service as a ClusterIP, NodePort, or LoadBalancer
ctrlPlane.advertisedHost	string	nil	global DNS name by which routers can resolve a reachable IP for this service: default is cluster service DNS name which assumes all routers are inside the same cluster
ctrlPlane.advertisedPort	int	443	cluster service, node port, load balancer, and ingress port
ctrlPlane.alternativeIssuer	string	nil	kind and name of alternative issuer for the controller's identity
ctrlPlane.containerPort	int	6262	cluster service target port on the container
ctrlPlane.dnsNames	list	[]	additional DNS SANs
ctrlPlane.ingress.annotations	string	nil	ingress annotations, e.g., to configure ingress-nginx
ctrlPlane.ingress.enabled	bool	false	create an ingress for the cluster service
ctrlPlane.service.enabled	bool	true	create a cluster service for the deployment
ctrlPlane.service.type	string	"ClusterIP"	expose the service as a ClusterIP, NodePort, or LoadBalancer
ctrlPlaneCasBundle.namespaceSelector	object	{}	namespaces where trust-manager will create the Bundle resource containing Ziti's trusted CA certs (default: empty means all namespaces)
dbFile	string	"ctrl.db"	name of the BoltDB file
edgeSignerPki.admin_client_cert.duration	string	"8760h"	admin client certificate duration as Go time.Duration
edgeSignerPki.admin_client_cert.renewBefore	string	"720h"	renew admin client certificate before expiry as Go time.Duration
edgeSignerPki.enabled	bool	true	generate a separate PKI root of trust for the edge signer CA
env	string	nil	set name to value in containers' environment
fabric.events.enabled	bool	false	enable fabric event logger and file handler
fabric.events.fileName	string	"fabric-events.json"
fabric.events.mountDir	string	"/var/run/ziti"
fabric.events.network.intervalAgeThreshold	string	"5s"	matching interval age and reporting interval ensures coherent metrics from fabric events
fabric.events.network.metricsReportInterval	string	"5s"	matching interval age and reporting interval ensures coherent metrics from fabric events
fabric.events.subscriptions[0].type	string	"fabric.circuits"
fabric.events.subscriptions[1].type	string	"fabric.links"
fabric.events.subscriptions[2].type	string	"fabric.routers"
fabric.events.subscriptions[3].type	string	"fabric.terminators"
fabric.events.subscriptions[4].metricFilter	string	".*"
fabric.events.subscriptions[4].sourceFilter	string	".*"
fabric.events.subscriptions[4].type	string	"metrics"
fabric.events.subscriptions[5].type	string	"edge.sessions"
fabric.events.subscriptions[6].type	string	"edge.apiSessions"
fabric.events.subscriptions[7].type	string	"fabric.usage"
fabric.events.subscriptions[7].version	int	3
fabric.events.subscriptions[8].type	string	"services"
fabric.events.subscriptions[9].interval	string	"5s"
fabric.events.subscriptions[9].type	string	"edge.entityCounts"
highAvailability.mode	string	"standalone"	Ziti controller HA mode
highAvailability.replicas	int	1	Ziti controller HA swarm replicas
image.additionalArgs	list	[]	additional arguments can be passed directly to the container to modify ziti runtime arguments
image.args	list	["{{ include \"configMountDir\" . }}/ziti-controller.yaml"]	args for the entrypoint command
image.command	list	["ziti","controller","run"]	container entrypoint command
image.homeDir	string	"/home/ziggy"	homeDir for admin login shell must align with container image's ~/.bashrc for ziti CLI auto-complete to work
image.pullPolicy	string	"IfNotPresent"	deployment image pull policy
image.repository	string	"docker.io/openziti/ziti-controller"	container image repository for app deployment
image.tag	string	""	override the container image tag specified in the chart
ingress-nginx.controller.extraArgs.enable-ssl-passthrough	string	"true"	configure subchart ingress-nginx to enable the pass-through TLS feature
ingress-nginx.enabled	bool	false	install the ingress-nginx subchart
managementApi.advertisedHost	string	nil	global DNS name by which routers can resolve a reachable IP for this service
managementApi.advertisedPort	int	443	cluster service, node port, load balancer, and ingress port
managementApi.containerPort	int	1281	cluster service target port on the container
managementApi.dnsNames	list	[]	additional DNS SANs
managementApi.ingress.annotations	string	nil	ingress annotations, e.g., to configure ingress-nginx
managementApi.ingress.enabled	bool	false	create an ingress for the cluster service
managementApi.service.enabled	bool	false	create a cluster service for the deployment
managementApi.service.type	string	"ClusterIP"	expose the service as a ClusterIP, NodePort, or LoadBalancer
network.createCircuitRetries	int	2	createCircuitRetries controls the number of retries that will be attempted to create a path (and terminate it) for new circuits.
network.cycleSeconds	int	15	Defines the period that the controller re-evaluates the performance of all of the circuits running on the network.
network.initialLinkLatency	string	"65s"	Sets the latency of link when it's first created. Will be overwritten as soon as latency from the link is actually reported from the routers. Defaults to 65 seconds.
network.minRouterCost	int	10	Sets router minimum cost. Defaults to 10
network.pendingLinkTimeoutSeconds	int	10	pendingLinkTimeoutSeconds controls how long we'll wait before creating a new link between routers where there isn't an established link, but a link request has been sent
network.routeTimeoutSeconds	int	10	routeTimeoutSeconds controls the number of seconds the controller will wait for a route attempt to succeed.
network.routerConnectChurnLimit	string	"1m"	Sets how often a new control channel connection can take over for a router with an existing control channel connection Defaults to 1 minute
network.smart.rerouteCap	int	4	Defines the hard upper limit of underperforming circuits that are candidates to be re-routed. If smart routing detects 100 circuits that are underperforming, and smart.rerouteCap is set to 1, and smart.rerouteFraction is set to 0.02, then the upper limit of circuits that will be re-routed in this cycleSeconds period will be limited to 1.
network.smart.rerouteFraction	float	0.02	Defines the fractional upper limit of underperforming circuits that are candidates to be re-routed. If smart routing detects 100 circuits that are underperforming, and smart.rerouteFraction is set to 0.02, then the upper limit of circuits that will be re-routed in this cycleSeconds period will be limited to 2 (2% of 100).
nodeSelector	object	{}	deployment template spec node selector
persistence.VolumeName	string	nil	PVC volume name
persistence.accessMode	string	"ReadWriteOnce"	PVC access mode: ReadWriteOnce (concurrent mounts not allowed), ReadWriteMany (concurrent allowed)
persistence.annotations	object	{}	annotations for the PVC
persistence.enabled	bool	true	required: place a storage claim for the BoltDB persistent volume
persistence.existingClaim	string	""	A manually managed Persistent Volume and Claim Requires persistence.enabled=true. If defined, PVC must be created manually before volume will be bound.
persistence.size	string	"2Gi"	2GiB is enough for tens of thousands of entities, but feel free to make it larger
persistence.storageClass	string	nil	Storage class of PV to bind. By default it looks for the default storage class. If the PV uses a different storage class, specify that here.
podAnnotations	object	{}	annotations to apply to all pods deployed by this chart
podSecurityContext	object	{"fsGroup":2171}	deployment template spec security context
podSecurityContext.fsGroup	int	2171	the GID of the group that should own any files created by the container, especially the BoltDB file
prometheus.advertisedHost	string	""	DNS name to advertise in place of the default internal cluster name built from the Helm release name
prometheus.advertisedPort	int	443	cluster service, node port, load balancer, and ingress port
prometheus.containerPort	int	9090	cluster service target port on the container
prometheus.service.annotations	object	{}
prometheus.service.enabled	bool	false	create a cluster service for the deployment
prometheus.service.labels	object	{"app":"prometheus"}	extra labels for matching only this service, ie. serviceMonitor
prometheus.service.type	string	"ClusterIP"	expose the service as a ClusterIP, NodePort, or LoadBalancer
prometheus.serviceMonitor.annotations	object	{}	ServiceMonitor annotations
prometheus.serviceMonitor.enabled	bool	true	If enabled, and prometheus service is enabled, ServiceMonitor resources for Prometheus Operator are created
prometheus.serviceMonitor.interval	string	nil	ServiceMonitor scrape interval
prometheus.serviceMonitor.labels	object	{}	Additional ServiceMonitor labels
prometheus.serviceMonitor.metricRelabelings	list	[]	ServiceMonitor relabel configs to apply to samples as the last step before ingestion https://github.com/prometheus-operator/prometheus-operator/blob/master/Documentation/api.md#relabelconfig (defines metric_relabel_configs)
prometheus.serviceMonitor.namespace	string	nil	Alternative namespace for ServiceMonitor resources
prometheus.serviceMonitor.namespaceSelector	object	{}	Namespace selector for ServiceMonitor resources
prometheus.serviceMonitor.relabelings	list	[]	ServiceMonitor relabel configs to apply to samples before scraping https://github.com/prometheus-operator/prometheus-operator/blob/master/Documentation/api.md#relabelconfig (defines relabel_configs)
prometheus.serviceMonitor.scheme	string	"https"	ServiceMonitor will use http by default, but you can pick https as well
prometheus.serviceMonitor.scrapeTimeout	string	nil	ServiceMonitor scrape timeout in Go duration format (e.g. 15s)
prometheus.serviceMonitor.targetLabels	list	[]	ServiceMonitor will add labels from the service to the Prometheus metric https://github.com/prometheus-operator/prometheus-operator/blob/main/Documentation/api.md#servicemonitorspec
prometheus.serviceMonitor.tlsConfig	object	{"insecureSkipVerify":true}	ServiceMonitor will use these tlsConfig settings to make the health check requests
prometheus.serviceMonitor.tlsConfig.insecureSkipVerify	bool	true	set TLS skip verify, because the SAN will not match with the pod IP
resources	object	{}	deployment container resources
securityContext	object	{}	deployment container security context
spireAgent.enabled	bool	false	if you are running a container with the spire-agent binary installed then this will allow you to add the hostpath necessary for connecting to the spire socket
spireAgent.spireSocketMnt	string	"/run/spire/sockets"	file path of the spire socket mount
tolerations	list	[]	deployment template spec tolerations
trust-manager.app.trust.namespace	string	"{{ .Release.Namespace }}"	trust-manager needs to be configured to trust the namespace in which the controller is deployed so that it will create the Bundle resource for the ctrl plane trust bundle
trust-manager.crds.enabled	bool	false	CRDs must be applied in advance of installing the parent chart
trust-manager.enabled	bool	false	install the trust-manager subchart
webBindingPki.enabled	bool	true	generate a separate PKI root of trust for web bindings, i.e., client, management, and prometheus APIs

TODO's

• replicas - Each controller replica needs to be it's own HA member. We have to wait until HA https://github.com/openziti/ziti/blob/release-next/doc/ha/overview.md is officially released.
• lower CA / Cert lifetime; how to refresh stuff when Certs are updated?
• Deploy Prometheus scraper configuration when prometheus.enabled = true
• cert-manager allows issuing only one cert per key, i.e., ClientCertKeyReuseIssue prevents us from issuing a user cert and server cert backed by same private key, hence the controller config.yaml re-uses server certs in place of user certs to allow startup and testing to continue