Deploy on Minikube

Deploying Orc8r on Minikube is the easiest way to test changes to the Helm charts. Most steps are similar to the main deployment guide, with a few differences.

Prerequisites

Spin up Minikube

Set up Minikube with one of the following commands, including sufficient resources and seeding the metrics config files:

• On macOS

  minikube start --cni=bridge --driver=hyperkit --memory=8gb --cpus=8 --kubernetes-version='v1.20.2' --mount --mount-string "${MAGMA_ROOT}/orc8r/cloud/docker/metrics-configs:/configs"

• On Linux

  minikube start --cni=bridge --driver=docker --memory=8gb --cpus=8 --kubernetes-version='v1.20.2' --mount --mount-string "${MAGMA_ROOT}/orc8r/cloud/docker/metrics-configs:/configs"

where the only difference between the two is the driver used.

Note:

• There are a lot of things that can go wrong with spinning up Minikube, so if you have problems check for documentation specific to your system.
• Make sure you are not connected to a VPN when running this command.
• For further information on the recommended drivers used above, see Minikube configs.

Now install prerequisites for Orc8r and create the orc8r K8s namespace

helm repo add bitnami https://charts.bitnami.com/bitnami
helm upgrade --install \
    --create-namespace \
    --namespace orc8r \
    --set global.postgresql.auth.postgresPassword=postgres,global.postgresql.auth.database=magma,fullnameOverride=postgresql \
    postgresql \
    bitnami/postgresql

Build and publish images

NOTE: skip this step if you want to use the official container images at linuxfoundation.jfrog.io.

There are 2 ways you can publish your own images: to a private registry, or to a localhost registry. Choose an option, then complete the relevant prerequisites:

1. Publish to private registry (specifically, we'll use DockerHub)
   • docker login
   • Use registry.hub.docker.com/USERNAME as your registry name
2. Publish to localhost registry (specifically, we'll use the local Docker registry via Minikube)
   • minikube addons enable registry
   • docker run --rm -it --network=host alpine ash -c "apk add socat && socat TCP-LISTEN:5000,reuseaddr,fork TCP:$(minikube ip):5000"
     • This should hang, open new tab for other commands
   • Use localhost:5000 as your registry name

After completing the prerequisites listed above, follow the instructions at Building Orchestrator to publish container images to the chosen registry.

Install

Generate secrets

If you haven't already, the easiest way to generate these secrets is temporarily spinning up a local, Docker-based deployment

export CERTS_DIR=${MAGMA_ROOT}/.cache/test_certs
cd ${MAGMA_ROOT}/orc8r/cloud/docker && ./build.py && ./run.py && sleep 30 && docker-compose down && ls -l ${CERTS_DIR} && cd -

Apply secrets

Create the K8s secrets

export IMAGE_REGISTRY_URL=linuxfoundation.jfrog.io/magma-docker  # or replace with your registry
export IMAGE_REGISTRY_USERNAME=''
export IMAGE_REGISTRY_PASSWORD=''

export CERTS_DIR=${MAGMA_ROOT}/.cache/test_certs  # mirrored from above

cd ${MAGMA_ROOT}/orc8r/cloud/helm/orc8r
helm template orc8r charts/secrets \
  --namespace orc8r \
  --set-string secret.certs.enabled=true \
  --set-file 'secret.certs.files.rootCA\.pem'=${CERTS_DIR}/rootCA.pem \
  --set-file 'secret.certs.files.bootstrapper\.key'=${CERTS_DIR}/bootstrapper.key \
  --set-file 'secret.certs.files.controller\.crt'=${CERTS_DIR}/controller.crt \
  --set-file 'secret.certs.files.controller\.key'=${CERTS_DIR}/controller.key \
  --set-file 'secret.certs.files.admin_operator\.pem'=${CERTS_DIR}/admin_operator.pem \
  --set-file 'secret.certs.files.admin_operator\.key\.pem'=${CERTS_DIR}/admin_operator.key.pem \
  --set-file 'secret.certs.files.certifier\.pem'=${CERTS_DIR}/certifier.pem \
  --set-file 'secret.certs.files.certifier\.key'=${CERTS_DIR}/certifier.key \
  --set-file 'secret.certs.files.nms_nginx\.pem'=${CERTS_DIR}/controller.crt \
  --set-file 'secret.certs.files.nms_nginx\.key\.pem'=${CERTS_DIR}/controller.key \
  --set=docker.registry=${IMAGE_REGISTRY_URL} \
  --set=docker.username=${IMAGE_REGISTRY_USERNAME} \
  --set=docker.password=${IMAGE_REGISTRY_PASSWORD} |
  kubectl apply -f -

Optional: fluentd secrets

To use fluentd on the minikube deployment, create additional fluentd secrets

cd ${CERTS_DIR}
openssl genrsa -out fluentd.key 2048
openssl req -new -key fluentd.key -out fluentd.csr -subj "/C=US/CN=fluentd.$domain"
openssl x509 -req -in fluentd.csr -CA certifier.pem -CAkey certifier.key -CAcreateserial -out fluentd.pem -days 3650 -sha256

Apply the secrets

cd ${MAGMA_ROOT}/orc8r/cloud/helm/orc8r
helm template orc8r charts/secrets \
  --namespace orc8r \
  --set-file 'secret.certs.files.fluentd\.pem'=${CERTS_DIR}/fluentd.pem \
  --set-file 'secret.certs.files.fluentd\.key'=${CERTS_DIR}/fluentd.key |
  kubectl apply -f -

Create values file

A minimal values file is at ${MAGMA_ROOT}/orc8r/cloud/helm/orc8r/examples/minikube.values.yaml

• Copy that file to ${MAGMA_ROOT}/orc8r/cloud/helm/orc8r.values.yaml
• Replace IMAGE_REGISTRY_URL with your registry and IMAGE_TAG with your tag
• Make additional edits as desired

Install charts

This section describes how to install based on local charts. However, you can also install charts from the official chart repositories

• Stable: https://linuxfoundation.jfrog.io/artifactory/magma-helm-prod/
• Test: https://linuxfoundation.jfrog.io/artifactory/magma-helm-test/

Install base orc8r chart

cd ${MAGMA_ROOT}/orc8r/cloud/helm/orc8r
helm dep update
helm upgrade --install --namespace orc8r --values ${MAGMA_ROOT}/orc8r/cloud/helm/orc8r.values.yaml orc8r .

Optionally install other charts, like the lte charts, similarly

cd ${MAGMA_ROOT}/lte/cloud/helm/lte-orc8r
helm dep update
helm upgrade --install --namespace orc8r --values ${MAGMA_ROOT}/orc8r/cloud/helm/lte.values.yaml lte .

Optionally install the fluentd charts by editing the file at ${MAGMA_ROOT}/orc8r/cloud/helm/orc8r/charts/logging/values.yaml

fluentd_daemon:
  create: true

  image:
    repository: IMAGE_REGISTRY_URL
    tag: IMAGE_TAG
    pullPolicy: IfNotPresent

  env:
    elastic_host: "host.minikube.internal"
    elastic_port: "9200"
    elastic_scheme: "http"

fluentd_forward:
  create: true

  # Domain-proxy output
  dp_output: true

  replicas: 1

  nodeSelector: {}
  tolerations: []
  affinity: {}

  image:
    repository: IMAGE_REPOSITORY
    tag: IMAGE_TAG
    pullPolicy: IfNotPresent

  env:
    elastic_host: "host.minikube.internal"
    elastic_port: "9200"
    elastic_scheme: "http"
    elastic_flush_interval: 5s

Replace IMAGE_REGISTRY_URL with your registry and IMAGE_TAG with your tag. Then install the charts

helm upgrade --install --namespace orc8r --values ${MAGMA_ROOT}/orc8r/cloud/helm/orc8r.values.yaml orc8r .

It may take a couple minutes before all pods are finished being created, but if successful, you should get something like this

$ kubectl --namespace orc8r get pods

NAME                                             READY   STATUS    RESTARTS   AGE
mysql-57955549d5-n69pd                           1/1     Running   0          6m14s
nms-magmalte-7c84667c4c-pvtlj                    1/1     Running   0          2m58s
nms-nginx-proxy-5b86f479f7-lvjpn                 1/1     Running   0          2m58s
orc8r-alertmanager-57d5d6ccc4-ht4n4              1/1     Running   0          2m58s
orc8r-alertmanager-configurer-76cf8f8f57-rmwjf   1/1     Running   0          2m58s
orc8r-controller-fdf59f456-vvqr2                 1/1     Running   0          2m58s
orc8r-nginx-7d6c78647-n6d8z                      1/1     Running   0          2m58s
orc8r-prometheus-77dccb799b-w9z6z                1/1     Running   0          2m58s
orc8r-prometheus-cache-6d647df4d9-2wqkc          1/1     Running   0          2m58s
orc8r-prometheus-configurer-6d6d987c88-pgm87     1/1     Running   0          2m58s
orc8r-user-grafana-6498bb6959-rchx5              1/1     Running   0          2m58s
postgresql-0                                     1/1     Running   4          6d23h

If using fluentd, there should be the pods orc8r-fluentd-forward and orc8r-fluentd-daemon as well

$ kubectl --namespace orc8r get pods

NAME                                             READY   STATUS    RESTARTS   AGE
mysql-57955549d5-n69pd                           1/1     Running   0          6m14s
nms-magmalte-7c84667c4c-pvtlj                    1/1     Running   0          2m58s
nms-nginx-proxy-5b86f479f7-lvjpn                 1/1     Running   0          2m58s
orc8r-alertmanager-57d5d6ccc4-ht4n4              1/1     Running   0          2m58s
orc8r-alertmanager-configurer-76cf8f8f57-rmwjf   1/1     Running   0          2m58s
orc8r-controller-fdf59f456-vvqr2                 1/1     Running   0          2m58s
orc8r-fluentd-forward-54794d4d86-599lv           1/1     Running   0          2m58s
orc8r-nginx-7d6c78647-n6d8z                      1/1     Running   0          2m58s
orc8r-prometheus-77dccb799b-w9z6z                1/1     Running   0          2m58s
orc8r-prometheus-cache-6d647df4d9-2wqkc          1/1     Running   0          2m58s
orc8r-prometheus-configurer-6d6d987c88-pgm87     1/1     Running   0          2m58s
orc8r-user-grafana-6498bb6959-rchx5              1/1     Running   0          2m58s
postgresql-0                                     1/1     Running   4          6d23h

$ kubectl --namespace kube-system get pods

NAME                               READY   STATUS    RESTARTS   AGE
coredns-74ff55c5b-xsz59            1/1     Running   0          3d
etcd-minikube                      1/1     Running   0          3d
kube-apiserver-minikube            1/1     Running   0          3d
kube-controller-manager-minikube   1/1     Running   0          3d
kube-proxy-cp6s5                   1/1     Running   0          3d
kube-scheduler-minikube            1/1     Running   0          3d
orc8r-fluentd-daemon-b7wpt         1/1     Running   0          4h
registry-46g6n                     1/1     Running   0          3d
registry-proxy-n6kml               1/1     Running   0          3d
storage-provisioner                1/1     Running   0          3d

Optionally, start the elasticsearch and kibana containers which handle the logs aggregated by fluentd

cd ${MAGMA_ROOT}/cloud/docker
./run.py

Access logs through Kibana

The Orchestrator logs aggregated by fluentd can be accessed via Kibana in the web browser under http://localhost:5601/. There should be two index patterns: fluentd (from fluentd-forward) and logstash (from fluentd-daemon). The logstash index pattern collects all logs from the Orchestrator, and fluentd the logs from connected gateways.

Configure

Access Orc8r

Create an Orc8r admin user

kubectl exec -it --namespace orc8r deploy/orc8r-orchestrator -- \
  /var/opt/magma/bin/accessc \
  add-existing -admin -cert /var/opt/magma/certs/admin_operator.pem \
  admin_operator

Now ensure the API and your certs are working

# Tab 1 (should hang)
kubectl --namespace orc8r port-forward svc/orc8r-nginx-proxy 7443:8443 7444:8444 9443:443

# Tab 2
export CERTS_DIR=${MAGMA_ROOT}/.cache/test_certs  # mirrored from above
curl \
  --insecure \
  --cert ${CERTS_DIR}/admin_operator.pem \
  --key ${CERTS_DIR}/admin_operator.key.pem \
  https://localhost:9443
# Should output: "hello"

Access NMS

Follow the instructions to create an NMS admin user

Port-forward Nginx

kubectl --namespace orc8r port-forward svc/nginx-proxy 8081:443

Log in to NMS at https://magma-test.localhost:8081 using credentials: admin@magma.test/password1234

Appendix

Minikube configs

We use the HyperKit driver on macOS instead of the default Docker driver because the former is more performant in supporting hairpin mode (via the CNI network plugin). I.e., Orc8r has some services that make requests to themselves, which requires K8s support.

You could also accomplish this with --cni=bridge --driver=docker, but this causes a 2x increase in pod startup times. Note that the HyperKit driver is not supported on Linux.