This section described how to deploy XOS. It is a minimal deployment that includes the set of micro-services that make up XOS itself, plus optionally, an example service (a web server running in a Kubernetes pod). For an example of how XOS is integrated into a larger system, see the CORD Installation Guide.
XOS runs on any version of Kubernetes (1.10 or greater), and uses the Helm client-side tool. If you are new to Kubernetes, we recommend this tutorial as a good place to start.
Although you are free to set up Kubernetes and Helm in whatever way makes sense to you, the following walks through an example installation sequence on MacOS. It was tested on version 10.12.6.
You need to install Docker. Visit
https://docs.docker.com/docker-for-mac/install/ for instructions.
You also need to install VirtualBox. Visit
https://www.virtualbox.org/wiki/Downloads for instructions.
The following assumes you've installed the Homebrew package manager. Visit
https://brew.sh/ for instructions.
Install Minikube and Kubectl
To install Minikube, run the following command:
curl -Lo minikube https://storage.googleapis.com/minikube/releases/v0.28.0/minikube-darwin-amd64 && chmod +x minikube && sudo mv minikube /usr/local/bin/
To install Kubectl, run the following command:
brew install kubectl
Install Helm and Tiller
The following installs both Helm and Tiller.
brew install kubernetes-helm
Bring Up a Kubernetes Cluster
Start a minikube cluster as follows. This automatically runs inside VirtualBox.
To see that it's running, type
You should see something like the following
Kubernetes master is running at https://192.168.99.100:8443 KubeDNS is running at https://192.168.99.100:8443/api/v1/namespaces/kube-system/services/kube-dns:dn s/proxy To further debug and diagnose cluster problems, use 'kubectl cluster-info dump'.
You can also see how the cluster is configured by looking at
Other tools described on this page use this configuration file to find your cluster.
If you want, you can see minikube running by looking at the VirtualBox dashboard. Or alternatively, you can visit the Minikube dashboard:
As a final setp, you need to start Tiller on the Kubernetes cluster.
Download XOS Helm-Charts
The helm charts used to deploy XOS are currently bundled in the CORD
helm-chart repository. The rest of this section assumes all you download
this repository into directory
mkdir $SRC_DIR cd $SRC_DIR git clone https://gerrit.opencord.org/helm-charts cd helm-charts
While downloading the simple
helm-charts repository is sufficient
for bringing up XOS, you may also want to download the XOS source
code, for example, so you can walk through the
XOS tutorial. The easiest way to do
this uses the
repo tool, as described here.
Bring Up XOS
xos-core (plus affiliated micro-services) into your
Kubernetes cluster, execute the following from the
helm dep update xos-core helm install xos-core -n xos-core
You also need to start the Kafka message bus to catch event notifications send by the various components:
helm repo add incubator http://storage.googleapis.com/kubernetes-charts-incubator helm install -f examples/kafka-single.yaml --version 0.8.8 -n cord-kafka incubator/kafka
kubectl get pods to verify that all containers that implement XOS
(and Kafka) are successfully running. You should see output that looks
something like this:
NAME READY STATUS RESTARTS AGE cord-kafka-0 1/1 Running 0 3m cord-kafka-zookeeper-0 1/1 Running 0 3m xos-chameleon-58c5b847d6-48jvf 1/1 Running 0 11m xos-core-7dc45f677b-pzhm6 1/1 Running 0 11m xos-db-c49549b7f-mzs4n 1/1 Running 0 11m xos-gui-7c96669d8c-8zkhq 1/1 Running 0 11m xos-tosca-7f6cf85657-gzddq 1/1 Running 0 11m xos-ws-5f47ff7d94-xg9f5 1/1 Running 0 11m
Bring Up Monitoring and Logging
Although not required, we recommend that you also bring up two auxilary services to capture and display monitoring and logging information. This is done by executing the following Helm charts.
Monitoring (once running, access Grafana Dashboard at port 31300):
helm dep update nem-monitoring helm install -n nem-monitoring nem-monitoring
Logging (once running, access Kibana Dashboard at port 30601):
helm dep up logging helm install -f examples/logging-single.yaml -n logging logging
-f examples/logging-single.yamloption says to not use persistent storage, which is fine for development or demo purposes, but not for operational deployments.
Bring Up a Service
Optionally, you can bring up a simple service to be managed by XOS.
This involves deploying two additional helm charts:
demo-simpleexampleservice. Again from the
directory, execute the following:
helm dep update xos-profiles/base-kubernetes helm install xos-profiles/base-kubernetes -n base-kubernetes helm dep update xos-profiles/demo-simpleexampleservice helm install xos-profiles/demo-simpleexampleservice -n demo-simpleexampleservice
When all the containers are successfully up and running,
pod will return output that looks something like this:
NAME READY STATUS RESTARTS AGE base-kubernetes-kubernetes-75d68b65bc-h594m 1/1 Running 0 6m base-kubernetes-tosca-loader-ltdzg 0/1 Completed 4 6m cord-kafka-0 1/1 Running 1 15m cord-kafka-zookeeper-0 1/1 Running 0 15m demo-simpleexampleservice-cc8fbfb7-s4r68 1/1 Running 0 5m demo-simpleexampleservice-tosca-loader-46qtg 0/1 Completed 4 5m xos-chameleon-58c5b847d6-rcqff 1/1 Running 0 16m xos-core-7dc45f677b-27vc9 1/1 Running 0 16m xos-db-c49549b7f-589n6 1/1 Running 0 16m xos-gui-7c96669d8c-gcwsv 1/1 Running 0 16m xos-tosca-7f6cf85657-bf276 1/1 Running 0 16m xos-ws-5f47ff7d94-mpn7g 1/1 Running 0 16m
tosca-loader items with
Completed status are jobs, as
opposed to pods. Their job is to load TOSCA-based provisioning and
configuration information into XOS, and so they run to complettion and
then terminate. It is not uncommon to see them in an
Error state as
they retry while waiting for the corresponding services to come
Visit XOS Dashboard
Finally, to view the XOS dashboard, run the following:
minikube service xos-gui
This will launch a window in your default browser. Administrator login
and password are defined in