4장. 자동 스케일링에 오케스트레이션 서비스(heat) 사용

프로세스

1. openstackclient Pod에 액세스합니다.

   $ oc rsh openstackclient

2. 템플릿의 디렉터리를 생성합니다(예: /tmp/templates/:)

   $ mkdir -p /tmp/templates

3. 인스턴스 구성 템플릿을 생성합니다(예: /tmp/templates/instance.yaml ):

4. instance.yaml 파일에 다음 구성을 추가합니다.

   cat <<EOF > /tmp/templates/instance.yaml
   heat_template_version: wallaby
   description: Template to control scaling of VNF instance
   
   parameters:
     metadata:
       type: json
     image:
       type: string
       description: image used to create instance
       default: cirros
     flavor:
       type: string
       description: instance flavor to be used
       default: m1.small
     network:
       type: string
       description: project network to attach instance to
       default: private
     external_network:
       type: string
       description: network used for floating IPs
       default: public
   
   resources:
     vnf:
       type: OS::Nova::Server
       properties:
         flavor: {get_param: flavor}
         image: { get_param: image }
         metadata: { get_param: metadata }
         networks:
           - port: { get_resource: port }
   
     port:
       type: OS::Neutron::Port
       properties:
         network: {get_param: network}
         security_groups:
           - basic
   
     floating_ip:
       type: OS::Neutron::FloatingIP
       properties:
         floating_network: {get_param: external_network }
   
     floating_ip_assoc:
       type: OS::Neutron::FloatingIPAssociation
       properties:
         floatingip_id: { get_resource: floating_ip }
         port_id: { get_resource: port }
   EOF

5. heat 템플릿에서 참조할 리소스를 생성합니다.

   $ cat <<EOF > /tmp/templates/resources.yaml
   resource_registry:
     "OS::Nova::Server::VNF": /tmp/templates/instance.yaml
   EOF

6. 오케스트레이션 서비스에 대한 배포 템플릿을 생성하여 인스턴스 스케일링을 제어합니다.

   cat <<EOF > /tmp/templates/autoscaling.yaml
   heat_template_version: wallaby
   description: Example auto scale group, policy and alarm
   resources:
     autoscalinggroup:
       type: OS::Heat::AutoScalingGroup
       properties:
         cooldown: 300
         desired_capacity: 1
         max_size: 3
         min_size: 1
         resource:
           # Configure the resource to be autoscaled
           type: OS::Nova::Server::VNF
           properties:
             metadata: {"metering.server_group": {get_param: "OS::stack_id"}}
   
     scaleup_policy:
       type: OS::Heat::ScalingPolicy
       properties:
         adjustment_type: change_in_capacity
         auto_scaling_group_id: { get_resource: autoscalinggroup }
         cooldown: 300
         scaling_adjustment: 1
   
     scaledown_policy:
       type: OS::Heat::ScalingPolicy
       properties:
         adjustment_type: change_in_capacity
         auto_scaling_group_id: { get_resource: autoscalinggroup }
         cooldown: 300
         scaling_adjustment: -1
   
     cpu_alarm_high:
       type: OS::Aodh::PrometheusAlarm
       properties:
         description: Scale up if CPU > 80%
         threshold: 80 # 80%
         comparison_operator: gt
         alarm_actions:
           - str_replace:
               template: trust+url
               params:
                 url: {get_attr: [scaleup_policy, signal_url]}
         query:
           str_replace:
             # ceilometer_cpu metric is in ns. Divide the rate by 10000000 to get percentage
             # The time duration in [] should be higher than ceilometer polling interval
             # You can add {ceilometer polling} to {prometheus scrape interval} to calculate the value.
             # The default value is 150s.
             template: "(rate(ceilometer_cpu{server_group=~'stack_id'}[150s]))/10000000"
             params:
               stack_id: {get_param: OS::stack_id}
   
     cpu_alarm_low:
       type: OS::Aodh::PrometheusAlarm
       properties:
         description: Scale up if CPU < 20%
         threshold: 20 # 20%
         comparison_operator: lt
         alarm_actions:
           - str_replace:
               template: trust+url
               params:
                 url: {get_attr: [scaledown_policy, signal_url]}
         query:
           str_replace:
             # ceilometer_cpu metric is in ns. Divide the rate by 10000000 to get percentage
             # The time duration in [] should be higher than ceilometer polling interval
             # You can add {ceilometer polling} to {prometheus scrape interval} to calculate the value.
             # The default value is 150s.
             template: "(rate(ceilometer_cpu{server_group=~'stack_id'}[150s]))/10000000"
             params:
               stack_id: {get_param: OS::stack_id}
   
   outputs:
     scaleup_policy_signal_url:
       value: {get_attr: [scaleup_policy, signal_url]}
   
     scaledown_policy_signal_url:
       value: {get_attr: [scaledown_policy, signal_url]}
   EOF