Warning
The main branch of this repository contains work-in-progress development code for the upcoming release, and is not guaranteed to be stable or working. It uses the WIP services which have Docker images built rrom the main branch of their repos.The source for the latest release of this repo can be found at Releases.
A Helm chart to easily deploy the EdgeX IoT project on Kubernetes. Based on EdgeX Minnesota version.
The helm chart is based on the secure and non-secure versions of the EdgeX Docker deployment scripts. This helm chart can be used as a starting point for your own EdgeX deployment.
- Kubernetes cluster 1.24+
- Helm 3.7.0+
Before starting, make sure you have curl and openssl installed locally. These tools are needed to generate TLS assets and test the configuration.
These instructions also assume that your Kubernetes cluster has an installed ingress controller and an installed load balancer.
This Helm chart was uses the stock Ngnix ingress controller, and a MetalLB load balancer (with an assigned IP of 192.168.122.200). Adaptations may be needed if your cluster uses something different.
To install the Nginx ingress controller, follow the instructions at https://kubernetes.github.io/ingress-nginx/deploy/#quick-start
To install MetalLB, follow the instructions at https://metallb.universe.tf/installation/#installation-with-helm
This Helm chart also arbitrarily configures the ingress route to
respond to the hostname edgex
(which should be passed in using the TLS Server Name Identification (SNI) feature).
This can also be customized as well using the edgex.security.tlsHost
setting.
Install the EdgeX helm chart with a release name edgex-napa
You can install the helm chart 1 of 2 ways.
- Install the chart from by cloning the edgex-helm repository
- Install the chart from the tar zipped asset from a tag
If you want to clone the entire edgex-helm repository:
$ git clone https://github.com/edgexfoundry/edgex-helm.git
$ git checkout v3.0
$ kubectl create namespace edgex
$ helm install edgex-napa -n edgex .
If you are only interested in installing the helm chart as a standalone:
-
Navigate to the tagged asset by selecting the tag you desire to use
-
Click on Downloads for the tag
- Either right click and copy the url or click to download the
edgex-helm.tar.gz
file to save locally
If you're using a Linux variant, you can use the copied URL to download the file from your terminal
curl -o edgex-helm.tar.gz <url for file from tag>
You will then need to unpack the tar zipped file:
tar -xvf edgex-helm.tar.gz
You will see output similar to:
You can now change to the unzipped directory and install the helm chart.
cd helm
kubectl create namespace edgex
helm install edgex-napa -n edgex .
helm uninstall edgex-napa -n edgex
EdgeX on kubernetes using NodePort type to expose services by default. You can use ping command to test whether the EdgeX services start successfully.
The ping command format:
http://<ExternalIP>:<ExposedPort>/api/v3/ping
For example, the edgex-core-data ping command format:
curl http://localhost:59880/api/v3/ping
With a modern browser, navigate to http://<ExternalIP>:30400.
Use details see EdgeX UI README
This section will cover some features of the Helm chart, so that it may be properly configured.
The helm chart pulls a number of standard Docker images from Docker Hub. This may cause Docker pull limits to be exceeded. If you have a Docker hub account, you may create a Docker image pull secrets that allows for more generous pull limits.
To create a Docker image pull secret, create the Kubernetes namespace for EdgeX, and run the following command:
$ kubectl create namespace edgex
$ kubectl create secret docker-registry dockerhub --namespace <namespace> --docker-server=https://index.docker.io/v1/ --docker-username=<username> --docker-password=<password> --docker-email=<email-address>
On the helm
command-line, specify the secrets to be used for pull secrets, for example: --set imagePullSecrets="{ dockerhub }"
.
The default value of edgex.storage.useHostPath
is true
.
This setting causes EdgeX data volumes to be created under /mnt
on the host file system,
which is a reasonable choice for a single-node Kubernetes deployment only.
For all other scenarios, set edgex.storage.useHostPath
to false
,
and configure the following settings:
edgex:
storage:
useHostPath: false
nonSharedVolumesClassName: "TBD"
nonSharedVolumesAccessMode: "ReadWriteOnce"
sharedVolumesClassName: "TBD"
sharedVolumesAccessMode: "ReadWriteMany"
The EdgeX helm char has been tested with both ReadWriteOnce
and ReadWriteMany
for both types of volumes,
and has been compatibility tested with Rook-Ceph, OpenEBS, and the Rancher LocalPath provisioner.
The default value of edgex.security.enabled
is false
. This may change in the future.
To enable the security, add the command flag in the helm install: --set edgex.security.enabled=true
Setting edgex.security.enabled
to true
during installation (recommended)
will enable microservice-level authentication
for EdgeX peer-to-peer communcation
as well authentication to Redis, Consul, and the MQTT broker, if used.
In lieu of a standalone API gateway used by the snap- and docker-based EdgeX deployments, the security-enabled helm chart is coded against a standard Kubernetes NGINX-based ingress controller, and ingress routes are configured to require client-side TLS authentication.
The device USB camera by default is not enabled to run in the deployment, to start it in a pod,
add the following command flag in the helm install: --set edgex.replicas.device.usbcamera=1
.
Run the following two scripts to generate key material and install it into the cluster.
The scripts create a server-side CA so that curl can trust the server, a client-side CA so that Nginx can trust the client, a server-side TLS certificate for the Nginx server to present to the client, and a client-side TLS certificate for the curl client to present to the server.
./tlsgen.sh
./tlsinstall.sh
To install with security features enabled,
set the edgex.security.enabled
flag to true
during installation.
If EdgeX was previously installed on security disabled, uninstall the non-security one first. The helm chart is not coded to allow for dynamic switching in and out of secure mode.
helm install edgex-napa --set edgex.security.enabled=true -n edgex .
The following job is included at the root of the helm chart as create-proxy-user-job.yaml
.
In this example, edgexuser
should be replaced with a username of your own choosing,
and the image
should be updated to an appropriate release image,
such as edgexfoundry/security-proxy-setup:3.0.0
.
apiVersion: batch/v1
kind: Job
metadata:
name: security-proxy-setup
spec:
template:
spec:
affinity:
podAffinity:
requiredDuringSchedulingIgnoredDuringExecution:
- labelSelector:
matchLabels:
org.edgexfoundry.service: edgex-security-secretstore-setup
topologyKey: "kubernetes.io/hostname"
imagePullSecrets:
- name: dockerhub
automountServiceAccountToken: false
containers:
- name: security-proxy-setup
image: nexus3.edgexfoundry.org:10004/security-proxy-setup:3.0.0
imagePullPolicy: Always
command: ["/edgex-init/ready_to_run_wait_install.sh"]
args: ["/edgex/secrets-config", "proxy", "adduser", "--user", "edgexuser", "--useRootToken"]
envFrom:
- configMapRef:
name: edgex-common-variables
securityContext:
allowPrivilegeEscalation: false
readOnlyRootFilesystem: true
volumeMounts:
- mountPath: /edgex-init
name: edgex-init
- mountPath: /vault/config
name: vault-config
- mountPath: /tmp/edgex/secrets
name: edgex-secrets
restartPolicy: Never
securityContext:
runAsNonRoot: false
runAsUser: 0
runAsGroup: 0
volumes:
- name: edgex-init
persistentVolumeClaim:
claimName: edgex-init
- name: vault-config
persistentVolumeClaim:
claimName: vault-config
- name: edgex-secrets
persistentVolumeClaim:
claimName: edgex-secrets
To get a credential, first run the job and inspect the output:
$ kubectl apply -f create-proxy-user.job.yaml
$ kubectl get pods
$ kubectl logs security-proxy-setup-m78g
{"username":"edgexuser","password":"dZ4SX...redacted...fOT"}
After the passwords is saved away, be sure to delete the completed job:
$ kubectl delete job security-proxy-setup
Next, create a script to obtain a secret store token and run it:
username=edgexuser
password="dZ4SX...redacted...fOT"
vault_token=$(curl -s --resolve edgex:443:192.168.122.200 --cacert server-ca.pem --cert client.pem --key client.key -X POST -H "Content-Type: application/json" "https://edgex/vault/v1/auth/userpass/login/${username}" -d "{\"password\":\"${password}\"}" | jq -r '.auth.client_token')
id_token=$(curl -s --resolve edgex:443:192.168.122.200 --cacert server-ca.pem --cert client.pem --key client.key -H "Authorization: Bearer ${vault_token}" "https://edgex/vault/v1/identity/oidc/token/${username}" | jq -r '.data.token')
echo "${id_token}"
The output will be a JWT of the form:
eyJ.redacted.redacted
Finally, test with curl
.
Note the use of the special options to enable SNI,
client-side certificates,
and server-side certificate validation.
Replace <MetalLB IP>
below with the external IP
that is servicing the Kubernetes ingress controller.
Use the $id_token
above to authenticate at the microservice layer.
curl -iv --resolve edgex:443:<MetalLB IP> --cacert server-ca.pem --cert client.pem --key client.key -H"Authorization: Bearer ${id_token}" "https://edgex/core-data/api/v3/version"
If everything was done correctly, the output will look like:
... a bunch of diagnostics ...
* Connection #0 to host edgex left intact
{"apiVersion":"v3","version":"3.0.0-dev.137","serviceName":"core-data"}
Consul has its own authentication mechanism. EdgeX saves a privielged token away when it sets up Consul's ACL mechanism. Use the following boilerplate code to retrieve it:
kubectl exec -ti edgex-core-consul-5b5cb7bd85-p26wc -- sh -c 'cat $STAGEGATE_REGISTRY_ACL_MANAGEMENTTOKENPATH | jq -r ".SecretID"'
Replace edgex-core-consul-5b5cb7bd85-p26wc
with the specific pod instance of the running Consul.
The helm chart exposes EdgeX services as ClusterIP
services by default.
This can be changed by the expose.type.<...>
settings on a per-service basis.
Since the security flag enables the ingress rules for EdgeX microservices by default,
it is unlikely that directly exposing an EdgeX service on its own port will be necessary.
The feature edgex.features.enableHostPort
is set to false
by default.
If enabled, EdgeX services behave as the do on the snap- and docker-based implementations:
the map service ports to localhost or the external network interface,
as specified in the hostPortXXXBind settings:
hostPortInternalBind: 127.0.0.1
hostPortExternalBind: 0.0.0.0
- VMware China R&D Center https://mp.weixin.qq.com/s/ECdEkc9QdkVScn4Lvl_JJA
If you find a bug or want to request a new feature, please open a GitHub Issue