...
The Software CR is another component of the Binary Provisioning Agent(BPA). The Software CR will install the required software, drivers and perform software updates on the hosts that make up the kubernetes cluster after the BPA controller has successfully installed kubernetes on the host. The required software and versions are specified in a configmap and the configmap name is specified in the custom resource when it is being createdthe software CR under the master and worker sub sections respectively. Software in the master section will be installed on all master nodes in the cluster and software in the worker section will be installed on all worker nodes in the cluster. If the software version is not specified, the latest version in that repository will be installed.
When the Software CR is created and the Kubernetes cluster has successfully been installed, the BPA controller looks up the matching software custom resource (cluster name in the Software CR must match cluster name in the provisioning CR), gets the configmap name software list and then installs the software (and/or updates) that are defined in the specified host hosts via SSH. Draft It gets the IP address for each host from the IP address configmap created by the BPA controller when it was installing the cluster
Software CRD
| Code Block | ||
|---|---|---|
| ||
apiVersion: apiextensions.k8s.io/v1beta1
kind: CustomResourceDefinition
metadata:
name: software.bpa.akraino.org
spec:
group: bpa.akraino.org
names:
kind: software
listKind: softwarerList
plural: software
singular: software
shortNames:
- sw
scope: Namespaced
subresources:
status: {}
validation:
openAPIV3Schema:
properties:
apiVersion:
description:
type: string
kind:
description:
type: string
metadata:
type: object
spec:
type: object
status:
type: object
version: v1alpha1
versions:
- name: v1alpha1
served: true
storage: true |
...
| Code Block |
|---|
package v1alpha1
import (
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
)
// EDIT THIS FILE! THIS IS SCAFFOLDING FOR YOU TO OWN!
// NOTE: json tags are required. Any new fields you add must have json tags for the fields to be serialized.
// SoftwareSpec defines the desired state of Software
// +k8s:openapi-gen=true
type SoftwareSpec struct {
// INSERT ADDITIONAL SPEC FIELDS - desired state of cluster
// Important: Run "operator-sdk generate k8s" to regenerate code after modifying this file
// Add custom validation using kubebuilder tags: https://book-v1.book.kubebuilder.io/beyond_basics/generating_crd.html
SoftwareConfigMap string//MasterSoftware map[string][]interface{} `json:"masterSoftware,omitempty"`
MasterSoftware []interface{} `json:"masterSoftware,omitempty"`
WorkerSoftware []interface{} `json:"workerSoftware,omitempty"`
//WorkerSoftware map[string][]interface{} `json:"SoftwareConfigMapNameworkerSoftware,requiredomitempty"`
}
// SoftwareStatus defines the observed state of Software
// +k8s:openapi-gen=true
type SoftwareStatus struct {
// INSERT ADDITIONAL STATUS FIELD - define observed state of cluster
// Important: Run "operator-sdk generate k8s" to regenerate code after modifying this file
// Add custom validation using kubebuilder tags: https://book-v1.book.kubebuilder.io/beyond_basics/generating_crd.html
}
// +k8s:deepcopy-gen:interfaces=k8s.io/apimachinery/pkg/runtime.Object
// Software is the Schema for the softwares API
// +k8s:openapi-gen=true
// +kubebuilder:subresource:status
type Software struct {
metav1.TypeMeta `json:",inline"`
metav1.ObjectMeta `json:"metadata,omitempty"`
Spec SoftwareSpec `json:"spec,omitempty"`
Status SoftwareStatus `json:"status,omitempty"`
}
// +k8s:deepcopy-gen:interfaces=k8s.io/apimachinery/pkg/runtime.Object
// SoftwareList contains a list of Software
type SoftwareList struct {
metav1.TypeMeta `json:",inline"`
metav1.ListMeta `json:"metadata,omitempty"`
Items []Software `json:"items"`
}
func init() {
SchemeBuilder.Register(&Software{}, &SoftwareList{})
}
|
SoftwareConfigMapMasterSoftware : This corresponds to SoftwareConfigMapName to the masterSoftware section in the SW sample CR (see sample file below). The Software Config Map will contain . It has a list of software and versions to install and update for each node type (master or worker).interfaces containing the various software, if the version is specified, the item is a map corresponding to software name and version, if it is not the item is a string that has the software name
WorkerSoftware: It is also as described in the masterSoftware
Sample Software CR YAML files
| Code Block | ||
|---|---|---|
| ||
apiVersion: bpa.akraino.org/v1alpha1 kind: Software metadata: name: example-software labels: cluster: cluster-abcxyz owner: c1 name: example-software spec: SoftwareConfigMapNamemasterSoftware: software-sample-configmap |
Sample Software ConfigMap
| Code Block | ||
|---|---|---|
| ||
apiVersion: v1 kind: ConfigMap metadata: name: software-sample-configmap data- curl - htop - jq: master: | curl:""version: 1.5+dfsg-1ubuntu0.1 - htopmaven:"" tmux:"" golang-go:"v12.1"version: 3.3.9-3 workerworkerSoftware: | - curl:"" - htop:"" - tmux jq:"" - jq tmux:"" |
...
Cluster IP Address Configmap
In order for the BPA controller to install the software specified in the software CR via SSH, it would need to have saved the IP addresses of the hosts in that cluster. The BPA controller creates a configmap that stores the hosts in a cluster and their corresponding IP address. Below is an example of the configmap the BPA operator creates;
| Code Block | ||||
|---|---|---|---|---|
| ||||
apiVersion: v1
data:
MASTER_master-1: 192.168.30.20
WORKER_worker-1: 192.168.30.65
kind: ConfigMap
metadata:
creationTimestamp: "2019-09-24T21:03:32Z"
labels:
cluster: cluster-abc
name: cluster-abc-configmap
namespace: default
resourceVersion: "8519202"
selfLink: /api/v1/namespaces/default/configmaps/cluster-abc-configmap
uid: c426f1fc-df0e-11e9-a107-00219ba0c77a
|