Skip to content

Manual

Jump to bottom
Catherine edited this page Nov 16, 2019 · 15 revisions

Create a key

To create a key, simply input key name and public key into fields and submit. If you don't have a key already, use command ssh-keygen -f /path/to/your_key to generate one.

Create an image

Input image name, url and architecture into create image tab and submit, wait till the image status become available. Centos7 image is created with the environment deployment.

Create a flavor

With allinone installation, a flavor named m1.tiny is created already, you can create more with your desired number of cpu, memory and disk size. Only admin has the privilege.

Create a subnet

You can specify a name with valid network, netmask, gateway, start and end into fields and submit. Make sure no conflict among your fields. Admin can create public or private and specify a vlan number, normal user can only create internal subnet with a random vxlan id. For what is public or private vlan, refer to network section in operation guide. A default subnet with username is created with user creation.

Create a security group with rules

A default security group named username is created when user is created. You can modify it, delete it or create new ones. The default security group has icmp and port 22 opened.

Launch an instance

With all the above, it is straightforward to create an instance via web ui. if you are admin, you can create instance on public or private subnets directly, otherwise you can only create instance on your own internal subnets. To access the instance on internal subnets, you must create a gateway, and then create a floating ip to access it. You can also edit or view the instance by clicking instance ID. To get VNC password, you need to refresh the edit page.

Create a gateway

To create a gateway, you can specify a name or choose what kind of network the gateway wants to route, it can be public, private or both, if you don't know what it is, leave it blank.

Create a floating IP

After you create a gateway, you can choose a instance from the drop down menu and create a floating ip to it.

Create an OpenShift cluster

It is simple to create an OpenShift cluster, you need to input cluster name, base domain which determine the domain names of your access endpoint and APPs, therefore, the combination of ${cluster_name}.${base_domain} must be a valid domain name. Also, you must have a key, and have a redhat account registered to get a pull secret (https://cloud.redhat.com/openshift/install/metal/user-provisioned) before proceeding with the create button. The combination of ${cluster_name}.${base_domain} must be a valid domain name.

During the cluster creation, the status indicates the stage where the installation goes. You can observe the instances created in instances tab. With time, the instances named ocID-lb, bootstrap, master-N, workers-N will show up in series. The instance ocID-lb acts as load balancer and domain name server for the cluster, you can login it via its floating IP with the key specified before and see the log file /tmp/ocd.log.

Once the cluster status is marked as complete, you can get the credentials from ocID-lb:/opt/${cluster_name}/auth. The web console is https://console-openshift-console.apps.\${cluster_name}.\${base_domain}.

There are 3 ways to access the DNS records of the APPs in this cluster

  1. Ideally, if you are the owner of ${base_domain}, then in your DNS provider's website, you can create
  • An 'A' record with dns.${cluster_name}.${base_domain} pointing to the public floating IP of instance ocID-lb
  • An 'NS' record with ${cluster_name}.${base_domain} referring to dns.${cluster_name}.${base_domain}
  1. For temporary usage or testing purpose, you can modify file /etc/hosts in your working machine with these records for example:
$floating_ip     console-openshift-console.apps.${cluster_name}.${base_domain}
$floating_ip     oauth-openshift.apps.${cluster_name}.${base_domain}
$floating_ip     prometheus-k8s-openshift-monitoring.apps.${cluster_name}.${base_domain}
$floating_ip     grafana-openshift-monitoring.apps.${cluster_name}.${base_domain}
$floating_ip     alertmanager-main-openshift-monitoring.apps.${cluster_name}.${base_domain}
$floating_ip     downloads-openshift-console.apps.${cluster_name}.${base_domain}
$floating_ip     default-route-openshift-image-registry.apps.${cluster_name}.${base_domain}

If you have more apps created, then create more records accordingly and similarly.

  1. Alternatively, you can modify /etc/resolv.conf in your working machine with
nameserver $floating_ip

The ocID-lb is able to resolve the domain names both inside and outside of the cluster.

Note: the installation procedure uses your access cookie, so do not logout cloudland web console before the installation completes. To scale up/down more/less workers, click the cluster ID and input worker number in edit page and submit

Create a Glusterfs cluster

To create a Glusterfs cluster, simply specify name, flavor and key. Optionally you can associate it with an existing OpenShift cluster and input a number larger than 3 for works. If an OpenShift cluster is associated, the vms will be deployed into its same subnet so the openshift workers can access it directly.

To create a dynamic storage class, run

cat >nfs-pv.yaml <<EOF
apiVersion: v1
kind: PersistentVolume
metadata:
  name: nfs-pv
spec:
  capacity:
    storage: 100Gi
  accessModes:
  - ReadWriteMany
  nfs:
    path: /opt/$cluster_name/data
    server: 192.168.91.8
  persistentVolumeReclaimPolicy: Recycle
EOF
oc create -f nfs-pv.yaml

To make the storage class as default storage class, run

kubectl patch storageclass glusterfs -p '{"metadata": {"annotations":{"storageclass.kubernetes.io/is-default-class":"true"}}}'

Note: The Gluster service requires a newer version fuse than OpenShift CoreOS has, so when the volumes are created and pvc bounded, they are likely not able to mount and you have to logon one of the Gluster worker nodes and run

for vol in `gluster volume list`; do
    gluster volume set $vol ctime off
done
Clone this wiki locally