Table of Contents
- Requirements
- Clone the repository
- Starting the cluster
- Running a container
- Running a user defined pod
- Troubleshooting
- I cannot reach service IPs on the network.
- I cannot create a replication controller with replica size greater than 1! What gives?
- I changed Kubernetes code, how do I run it?
- kubectl claims to start a container but
get pods
anddocker ps
don't show it. - The pods fail to connect to the services by host names
Not running Linux? Consider running Minikube, or on a cloud provider like Google Compute Engine.
At least Docker
1.10+. Ensure the Docker daemon is running and can be contacted (try docker ps
). Some of the Kubernetes components need to run as root, which normally
works fine with docker.
You need an etcd in your path, please make sure it is installed and in your $PATH
.
You need go in your path (see here for supported versions), please make sure it is installed and in your $PATH
.
You need OpenSSL installed. If you do not have the openssl
command available, the script will print an appropriate error.
The CFSSL binaries (cfssl, cfssljson) must be installed and available on your $PATH
.
The easiest way to get it is something similar to the following:
$ go get -u github.com/cloudflare/cfssl/cmd/...
$ PATH=$PATH:$GOPATH/bin
In order to run kubernetes you must have the kubernetes code on the local machine. Cloning this repository is sufficient.
$ git clone --depth=1 https://github.com/kubernetes/kubernetes.git
The --depth=1
parameter is optional and will ensure a smaller download.
In a separate tab of your terminal, run the following (since one needs sudo access to start/stop Kubernetes daemons, it is easier to run the entire script as root):
cd kubernetes
hack/local-up-cluster.sh
This will build and start a lightweight local cluster, consisting of a master and a single node. Type Control-C to shut it down.
If you've already compiled the Kubernetes components, then you can avoid rebuilding them with this script by using the -O
flag.
./hack/local-up-cluster.sh -O
You can use the cluster/kubectl.sh script to interact with the local cluster. hack/local-up-cluster.sh will print the commands to run to point kubectl at the local cluster.
Your cluster is running, and you want to start running containers!
You can now use any of the cluster/kubectl.sh commands to interact with your local setup.
cluster/kubectl.sh get pods
cluster/kubectl.sh get services
cluster/kubectl.sh get replicationcontrollers
cluster/kubectl.sh run my-nginx --image=nginx --replicas=2 --port=80
## begin wait for provision to complete, you can monitor the docker pull by opening a new terminal
sudo docker images
## you should see it pulling the nginx image, once the above command returns it
sudo docker ps
## you should see your container running!
exit
## end wait
## introspect Kubernetes!
cluster/kubectl.sh get pods
cluster/kubectl.sh get services
cluster/kubectl.sh get replicationcontrollers
Note the difference between a container
and a pod. Since you only asked for the former, Kubernetes will create a wrapper pod for you.
However you cannot view the nginx start page on localhost. To verify that nginx is running you need to run curl
within the docker container (try docker exec
).
You can control the specifications of a pod via a user defined manifest, and reach nginx through your browser on the port specified therein:
cluster/kubectl.sh create -f test/fixtures/doc-yaml/user-guide/pod.yaml
Congratulations!
Some firewall software that uses iptables may not interact well with
kubernetes. If you have trouble around networking, try disabling any
firewall or other iptables-using systems, first. Also, you can check
if SELinux is blocking anything by running a command such as journalctl --since yesterday | grep avc
.
By default the IP range for service cluster IPs is 10.0.. - depending on your docker installation, this may conflict with IPs for containers. If you find containers running with IPs in this range, edit hack/local-cluster-up.sh and change the service-cluster-ip-range flag to something else.
You are running a single node setup. This has the limitation of only supporting a single replica of a given pod. If you are interested in running with larger replica sizes, we encourage you to try the local vagrant setup or one of the cloud providers.
cd kubernetes
make
hack/local-up-cluster.sh
One or more of the Kubernetes daemons might've crashed. Tail the logs of each in /tmp.
To start the DNS service, you need to set the following variables:
KUBE_ENABLE_CLUSTER_DNS=true
KUBE_DNS_SERVER_IP="10.0.0.10"
KUBE_DNS_DOMAIN="cluster.local"
To know more on DNS service you can check out the docs.