Skip to content

Latest commit

 

History

History
412 lines (323 loc) · 17 KB

README.md

File metadata and controls

412 lines (323 loc) · 17 KB
Raw

Getting Started on vSphere

Prerequisites

  • If Kubernetes Cluster is deployed on vSphere Cluster, then make sure time is in sync on all esx hosts in the cluster otherwise deployment may fail due to certificate expiry check.
  • docker-engine for executing the kubernetes-anywhere deployment which can be downloaded here.
  • Kubernetes Anywhere is tested on vSphere deployments with vCenter, single node vSphere without vCenter is not supported.
  • Deployment requires DHCP server in the VM network.
  • vCenter user with following minimal set of privileges. 
    Datastore > Allocate space
Datastore > Low level file Operations
Folder > Create Folder
Folder > Delete Folder
Network > Assign network
Resource > Assign virtual machine to resource pool
Virtual machine > Configuration > Add new disk
Virtual Machine > Configuration > Add existing disk
Virtual Machine > Configuration > Add or remove device
Virtual Machine > Configuration > Change CPU count
Virtual Machine > Configuration > Change resource
Virtual Machine > Configuration > Memory
Virtual Machine > Configuration > Modify device settings
Virtual Machine > Configuration > Remove disk
Virtual Machine > Configuration > Rename
Virtual Machine > Configuration > Settings
Virtual machine > Configuration > Advanced
Virtual Machine > Interaction > Power off
Virtual Machine > Interaction > Power on
Virtual Machine > Inventory > Create from existing
Virtual Machine > Inventory > Create new
Virtual Machine > Inventory > Remove
Virtual Machine > Provisioning > Clone virtual machine
Virtual Machine > Provisioning > Customize
Virtual Machine > Provisioning > Read customization specifications
vApp > Import
Profile-driven storage -> Profile-driven storage view

Note: vSphere Cloud Provider doesn't need these many privileges. These privileges are required for deployment of Kubernetes Cluster using Kubernetes-Anywhere. Please refer vSphere Cloud Provider Guide for minimal set of privileges required for vSphere Cloud Provider.

Deployment

Note:

  • The recommended version is v1.6.5
  • The deployment is tested with kubernetes 1.6.5, 1.5.7
  • vSphere Cloud Provider is tested on v1.6.5, v1.5.7

Upload VM image to be used to vSphere:

Upload the template OS OVA to vCenter before deploying kubernetes. All kubernetes nodes will be clones of this VM.

Upload using vSphere Client.

  1. Login to vSphere Client.
  2. Right-Click on ESX host on which you want to deploy template.
  3. Select Deploy OVF template.
  4. Copy and paste URL for OVA For vSphere 6.0 and above (Updated on March 1 2017). OVA For vSphere 5.5 with Virtual Machine Hardware Version 10
  5. Check the name of the VM created, this will be used to deploy kubernetes later. (Should default to KubernetesAnywhereTemplatePhotonOS.ova)

You can also upload ova using govc. This OVA is based on Photon OS(v1.0) with virtual hardware v11.

NOTE: DO NOT POWER ON THE IMPORTED VM.

If you do power it on, future clones of this VM will end up getting the same IP as the imported VM. To work around this run the following command before powering the VM off and using it to clone the kubernetes nodes.

echo -n > /etc/machine-id

Launch deployment environment

Pull cnastorage/kubernetes-anywhere:latest docker image:

docker pull cnastorage/kubernetes-anywhere

Run docker container using cnastorage/kubernetes-anywhere:latest image to launch deployment environment:

docker run -it -v /tmp:/tmp --rm --env="PS1=[container]:\w> " --net=host cnastorage/kubernetes-anywhere:latest /bin/bash

Note: Here in the above command we are mounting local /tmp directory, so that after the deployment is finished successfully, we can copy kubeconfig.json file on local system from the deployment container.

Start the deployment wizard:

Sample config

Lets take a look at vSphere environment before starting deployment wizard.

k8s-vsphere-deployment-01

Here you see Datacenter PA-DC, has cluster-vsan-1 with one resource pool - dev-resource-pool.

Following steps will deploy 4 nodes Kubernetes Cluster on the dev-resource-pool using the template file deployed on the cluster.

Let's start the deployment wizard. on the container prompt execute make config from /opt/kubernetes-anywhere directory

[container]:/opt/kubernetes-anywhere> make config

and complete the config wizard to create configuration for the kubernetes cluster. You can get help for any config option by entering '?'.

  • Select the number of nodes. Master + Number of nodes will be deployed.
number of nodes (phase1.num_nodes) [4] (NEW) 4
  • Set the cluster name. A folder with the cluster name will be created to place all the VMs.
cluster name (phase1.cluster_name) [kubernetes] (NEW) kubernetes
  • phase1.ssh_user - This field is not used for vSphere Deployment. Leave it blank and hit enter.
SSH user to login to OS for provisioning (phase1.ssh_user) [] (NEW)
  • Select the provider, in this case it would be vsphere.
cloud provider: gce, azure or vsphere (phase1.cloud_provider) [vsphere] (NEW) vsphere
  • Set the vCenter URL (Just the IP or domain name, without https://)
  vCenter URL Ex: 10.192.10.30 or myvcenter.io (phase1.vSphere.url) [] (NEW) 10.160.0.77
  • Set the port for vCenter communication. Unless vCenter is setup with a different port select the default port.
  vCenter port (phase1.vSphere.port) [443] (NEW)
  • Enter the user name for vCenter. All vCenter operations will be performed using these user credentials
  vCenter username (phase1.vSphere.username) [] (NEW) administrator@myvcenter.io
  • Enter the password for vCenter.
  vCenter password (phase1.vSphere.password) [] (NEW) MyPassword#3
  • Set the type of certificate used by vCenter. Set to true for self signed certificates
  Does host use self-signed cert (phase1.vSphere.insecure) [Y/n/?] (NEW) Y
  • Set the datacenter in vCenter to use. Specify the same datacenter to which the OVA was imported to.
  Datacenter (phase1.vSphere.datacenter) [datacenter] (NEW) cna-storage
  • Set the datastore to be use. This will be used for placing the VMs and volumes created via storage classes/dynamic provisioning.
  Datastore (phase1.vSphere.datastore) [datastore] (NEW) vsanDatastore
  • Specify where to deploy Kubernetes Cluster on Host or on Cluster.
   Deploy Kubernetes Cluster on 'host' or 'cluster' (phase1.vSphere.placement) [cluster] (NEW) cluster
  • If Host is selected, Specify host IP or FQDN. If Cluster is selected, Specify cluster name
    vsphere cluster name. Please make sure that all the hosts in the cluster are time-synchronized otherwise some of the nodes can remain in pending state for ever due to expired certificate (phase1.vSphere.cluster) [] (NEW) cluster-vsan-1
  • Specify yes if Kubernetes Cluster needs to be deployed on the resource pool located in selected host or cluster
    Do you want to use the existing resource pool created on the host or cluster? [yes, no] (phase1.vSphere.useresourcepool) [no] (NEW) yes
  • Specify name of the resource pool.
  Name of the Resource Pool. If Resource pool is enclosed within another Resource pool, specify pool hierarchy as ParentResourcePool/ChildResourcePool (phase1.vSphere.resourcepool) (NEW) dev-resource-pool
  • Specify the folder name or folder path where Kubernetes Node VMs should placed in the VC Inventory. Folder path will be created if not present.
  VM Folder name or Path (e.g kubernetes, VMFolder1/dev-cluster, VMFolder1/Test Group1/test-cluster). Folder path will be created if not present (phase1.vSphere.vmfolderpath) [kubernetes] (NEW) kubernetes
  • Number of vCPUs for each VM. Master and all nodes will have the number of vCPUs configured below.
  Number of vCPUs for each VM (phase1.vSphere.vcpu) [1] (NEW) 4
  • Memory for each VM. Master and all nodes will have the RAM configured below.
  Memory for each VM (phase1.vSphere.memory) [2048] (NEW) 4096
  • Network for each VM. Master and all nodes will have the Network configured below.
  Network for each VM (phase1.vSphere.network) [VM Network] (NEW)
  • Name of the template VM to use to create clone VMs for master and all nodes. The name here should be the same as the name that is reported by vCenter.
  Name of the template VM imported from OVA. If Template file is not available at the destination location specify vm path (phase1.vSphere.template) [KubernetesAnywhereTemplatePhotonOS.ova] (NEW) /PA-DC/vm/KubernetesAnywhereTemplatePhotonOS
  • Configure the POD network using flannel
  Flannel Network (phase1.vSphere.flannel_net) [172.1.0.0/16] (NEW)
  • Ignition image to be used for phase 2. For Kubernetes release 1.6 and above use docker.io/cnastorage/k8s-ignition:v2. For older releasees use docker.io/cnastorage/k8s-ignition:v1
*
* Phase 2: Node Bootstrapping
*
installer container (phase2.installer_container) [docker.io/cnastorage/k8s-ignition:v2] (NEW) docker.io/cnastorage/k8s-ignition:v2
  • Registry to be used by Kubernetes
docker registry (phase2.docker_registry) [gcr.io/google-containers] (NEW)
  • Set the release of Kubernetes to be used. The release should be the exact string used to tag a release.
kubernetes version (phase2.kubernetes_version) [v1.6.5] (NEW) v1.6.5
  • Set bootstrap provider to ignition
bootstrap provider (phase2.provider) [ignition] (NEW) ignition
  • Select the addons. Defaults to yes. Make sure to choose 'N' for weave-net addon.
*
* Phase 3: Deploying Addons.
*
Run the addon manager? (phase3.run_addons) [Y/n/?] (NEW)
  Run kube-proxy? (phase3.kube_proxy) [Y/n/?] (NEW)
  Run the dashboard? (phase3.dashboard) [Y/n/?] (NEW)
  Run heapster? (phase3.heapster) [Y/n/?] (NEW)
  Run kube-dns? (phase3.kube_dns) [Y/n/?] (NEW)
  Run weave-net? (phase3.weave_net) [N/y/?] (NEW) N

#
# configuration written to .config
#

Start the deployment using .config file

[container]:/opt/kubernetes-anywhere> make deploy
util/config_to_json /opt/kubernetes-anywhere/.config > /opt/kubernetes-anywhere/.config.json
make do WHAT=deploy-cluster
make[1]: Entering directory '/opt/kubernetes-anywhere'
.
.
.
KUBECONFIG="$(pwd)/phase1/vsphere/kubernetes/kubeconfig.json" ./util/validate
Validation: Expected 5 (workers + master) healthy nodes; found 0. (10s elapsed)
Validation: Expected 5 (workers + master) healthy nodes; found 2. (20s elapsed)
Validation: Expected 5 (workers + master) healthy nodes; found 5. (30s elapsed)
Validation: Success!
.
.
+ kubectl apply -f kubernetes/.tmp
deployment "kubernetes-dashboard" created
service "kubernetes-dashboard" created
deployment "heapster-v1.2.0" created
service "heapster" created
replicationcontroller "kube-dns-v19" created
service "kube-dns" created
daemonset "kube-proxy" created

Notes:

  • if OVA file is not located in the resource pool, where you want to deploy Kubernetes Cluster, please specify full VM Path.
  • You can build your own phase2.installer_container using Dockerfile here.
  • To change configuration, run: make config .config. Run make clean before make deploy

Congratulations!

You have a Kubernetes cluster!

Lets take a look at where node VM's are located on the vCenter.

Node VMs are placed in the dev-resource-pool under the VM Folder kubernetes. kubernetes is the name of the cluster we specified in the configuration wizard.

k8s-vsphere-deployment-02

k8s-vsphere-deployment-03

To deploy another kubernetes cluster while keeping the existing one, run make config and specify different kubernetes cluster name (phase1.cluster_name) and resource pool name and follow steps mentioned above.

Next Steps:

First set KUBECONFIG to access cluster using kubectl:

export KUBECONFIG="/opt/kubernetes-anywhere/phase1/vsphere/kubernetes/kubeconfig.json"

Note: In the path /opt/kubernetes-anywhere/phase1/vsphere/kubernetes kubernetes is the name of the cluster that we have specifed in the config file. If you have specified different name make sure to specify the appropriate path.

We have mounted /tmp directory in the deployment container. If you want to save the config on your local machine just copy this file to the /tmp directory in the container.

You will get cluster information when you run:

kubectl cluster-info

To access the dashboard after successful installation of kubernetes cluster. There are 2 options.

  • Run kubectl proxy outside the container created from cnastorage/kubernetes-anywhere:latest

Note: Make sure to download the kubectl version that matches with deployed kubernetes cluster

curl -LO https://storage.googleapis.com/kubernetes-release/release/v1.6.5/bin/linux/amd64/kubectl
chmod u+x kubectl
mkdir ~/.kube/
cd  ~/.kube/
vi config // copy content of $(make -s kubeconfig-path) and paste in this file.
export KUBECONFIG=~/.kube/config
./kubectl proxy
Starting to serve on 127.0.0.1:8001

Open the http://127.0.0.1:8001/ui in a browser

  • Access the dashboard from the node it is running on via NodePort mapping
# Get NodePort mapping
kubectl describe service kubernetes-dashboard --namespace=kube-system| grep -i NodePort
Type:                   NodePort
NodePort:               <unset> 31402/TCP

# Get node it is running on
kubectl get pods --namespace=kube-system| grep -i dashboard
  kubernetes-dashboard-1763797262-fzla9   1/1       Running   0          13m

kubectl describe pod kubernetes-dashboard-1763797262-fzla9 --namespace=kube-system| grep Node
  Node:           node1/10.20.104.220

# Select the public IP for the node via or use govc or vCenter UI
kubectl describe node node1| grep Address
# Open the <IP Addr>:<NodePort> in a browser

Destroy

After you've had a great experience with Kubernetes, run:

$ make destroy

to tear down your cluster.

If make destroy fails due to a known issue, the VMs can be deleted from vCenter.

Known Issues

  1. make destroy is flaky.
  2. Only a single kubernetes cluster can exist in a resource pool.

Troubleshooting

Logging into the VMs

The default password is not yet configurable, the default login is

user: root
password: kubernetes

Validation Fails (Zero nodes are healthy)

If no nodes are available, there was likely a provisioning failure on the master (either in vSphere or in the ignition provisioning container). The following steps will help in troubleshooting:

  1. SSH to the master.
  2. Use the following command to get relevant logs:
    • journalctl -u kubelet
  3. Attach the logs to a new Issue in this repository.

Validation Fails (One or more nodes are missing/unhealthy)

  1. Use kubectl get nodes to identify the missing nodes.
  2. Use vSphere Client or govc to find the node and the node's IP address.
  3. SSH to the master, then to the missing node
  4. Use the following command to get relevant logs:
    • journalctl -u kubelet
  5. Attach the logs to a new Issue in this repository.

Validation Fails (Dashboard or other kubernetes services are not working)

This was be mostly likely flannel failure.

  1. Use kubectl describe pod dashboard-pod-name to identify the node on which dashboard pod is scheduled.
  2. Use vSphere Client or govc to find the node and the node's IP address.
  3. SSH to the node.
  4. Use the following command on node to get relevant logs:
    • journalctl -u flannelc
  5. Attach the logs to a new Issue in this repository.