Beijing Scope

During the Beijing release the following capabilities are added or enhanced from the OOM capabilities available in the Amsterdam release:

• Deploy - with built-in component dependency management (including multiple clusters, federated deployments across sites, and anti-affinity rules)
• Configure - unified configuration across all ONAP components
• Monitor - real-time health monitoring feeding to a Consul UI and Kubernets
• Heal- failed ONAP containers are recreated automatically
• Scaling - cluster ONAP services to enable seamless scaling 
• Upgrade - change-out containers or configuration with little or no service impact
• Delete - cleanup individual containers or entire deployments

The following sections describe these areas in more detail.

Deploy

The OOM team with assistance from the ONAP project teams, have built a comprehensive set of Kubernetes deployment specifications, yaml files very similar to TOSCA files, that describe the composition of each of the ONAP components and the relationship within and between components. These deployment specifications describe the desired state of an ONAP deployment and instruct the Kubernetes container manager as to how to maintain the deployment in this state.  These dependencies dictate the order in-which the containers are started for the first time such that such dependencies are always met without arbitrary sleep times between container startups.  For example, the SDC back-end container requires the Elastic-Search, Cassandra and Kibana containers within SDC to be ready and is also dependent on DMaaP (or the message-router) to be ready - where ready implies the built-in "readiness" probes succeeded - before becoming fully operational. 

When an initial deployment of ONAP is requested the current state of the system is NULL so ONAP is deployed by the Kubernetes manager as a set of Docker containers on one or more predetermined hosts.  The hosts could be physical machines or virtual machines.  When deploying on virtual machines the resulting system will be very similar to “Heat” based deployments, i.e. Docker containers running within a set of VMs, the primary difference being that the allocation of containers to VMs is done dynamically with OOM and statically with “Heat”.

Example SO deployment descriptor file shows SO's dependency on its mariadb component:

..
kind: Deployment
metadata:
  name: mso
..
spec:
..

      initContainers:
      - command:
        - /root/ready.py
        args:
        - --container-name
        - mariadb
..        
      containers:
      - command:
        - /tmp/start-jboss-server.sh
        image: {{ .Values.image.mso }}
        imagePullPolicy: {{ .Values.pullPolicy }}
        name: mso
..

Pod Placement Rules

OOM will use the rich set of Kubernetes node and pod affinity / anti-affinity rules to minimize the chance of a single failure resulting in a loss of ONAP service. Node affinity / anti-affinity is used to guide the Kubernetes orchestrator in the placement of pods on nodes (physical or virtual machines).  For example:

• if a container used Intel DPDK technology the pod may state that it as affinity to an Intel processor based node, or
• geographical based node labels (such as the Kubernetes standard zone or region labels) may be used to ensure placement of a DCAE complex close to the VNFs generating high volumes of traffic thus minimizing networking cost. Specifically, if nodes were pre-assigned labels East and West, the pod deployment spec to distribute pods to these nodes would be: 

nodeSelector:
    failure-domain.beta.kubernetes.io/region: {{ .Values.location }}

Where "location: West" is specified in the values.yaml file used to deploy one DCAE cluster and  "location: East" is specified in a second values.yaml file (see OOM Configuration Management for more information about configuration files like the values.yamlfile).

Node affinity can also be used to achieve geographic redundancy if pods are assigned to multiple failure domains. For more information refer to Assigning Pods to Nodes.  Kubernetes has a comprehensive system called Taints and Tolerations that can be used to force the container orchestrator to repel pods from nodes based on static events (an administrator assigning a taint to a node) or dynamic events (such as a node becoming unreachable or running out of disk space). There are no plans to use taints or tolerations in the ONAP Beijing release.

Pod affinity / anti-affinity is the concept of creating a spacial relationship between pods when the Kubernetes orchestrator does assignment (both initially an in operation) to nodes as explained in Inter-pod affinity and anti-affinity. For example, one might choose to co-located all of the ONAP SDC containers on a single node as they are not critical runtime components and co-location minimizes overhead. On the other hand, one might choose to ensure that all of the containers in an ODL cluster (SDNC and APPC) are placed on separate nodes such that a node failure has minimal impact to the operation of the cluster.  An example of how pod affinity / anti-affinity is shown below:

apiVersion: v1
kind: Pod
metadata:
  name: with-pod-affinity
spec:
  affinity:
    podAffinity:
      requiredDuringSchedulingIgnoredDuringExecution:
      - labelSelector:
          matchExpressions:
          - key: security
            operator: In
            values:
            - S1
        topologyKey: failure-domain.beta.kubernetes.io/zone
    podAntiAffinity:
      preferredDuringSchedulingIgnoredDuringExecution:
      - weight: 100
        podAffinityTerm:
          labelSelector:
            matchExpressions:
            - key: security
              operator: In
              values:
              - S2
          topologyKey: kubernetes.io/hostname
  containers:
  - name: with-pod-affinity
    image: gcr.io/google_containers/pause:2.0

This example contains both podAffinity and podAntiAffinity rules, the first rule is is a must (requiredDuringSchedulingIgnoredDuringExecution) while the second will be met pending other considerations (preferredDuringSchedulingIgnoredDuringExecution).

Preemption

Another feature that may assist in achieving a repeatable deployment in the presence of faults that may have reduced the capacity of the cloud is assigning priority to the containers such that mission critical components have the ability to evict less critical components.  Kubernetes provides this capability with Pod Priority and Preemption.

Prior to having more advanced production grade features available, the ability to at least be able to re-deploy ONAP (or a subset of) reliably provides a level of confidence that should an outage occur the system can be brought back on-line predictably.

The overall Epic for OOM based deployment of ONAP is  OOM-6 - Getting issue details... STATUS  with the following status: 

During the Beijing release the 'initContainers' constructs were updated from the previous beta implementation to the current released syntax under the JIRA Story  OOM-406 - Getting issue details... STATUS

Configure

Each project within ONAP has its own configuration data generally consisting of: environment variables, configuration files, and database initial values.  Many technologies are used across the projects resulting in significant operational complexity and an inability to apply global parameters across the entire ONAP deployment. OOM solves this problem by introducing a common configuration technology, Helm charts, that provide a hierarchical configuration configuration with the ability to override values with higher level charts or command line options.  For example, if one wishes to change the OpenStack instance oam_network_cidr and ensure that all ONAP components reflect this change, one could change the vnfDeployment/openstack/oam_network_cidr value in the global configuration file as shown below:

nsPrefix: onap
nodePortPrefix: 302
apps: consul msb mso message-router sdnc vid robot portal policy appc aai sdc dcaegen2 log cli multicloud clamp vnfsdk aaf kube2msb
dataRootDir: /dockerdata-nfs

# docker repositories
repository:
  onap: nexus3.onap.org:10001
  oom: oomk8s
  aai: aaionap
  filebeat: docker.elastic.co

image:
  pullPolicy: Never

# vnf deployment environment
vnfDeployment:
  openstack:
    ubuntu_14_image: "Ubuntu_14.04.5_LTS"
    public_net_id: "e8f51956-00dd-4425-af36-045716781ffc"
    oam_network_id: "d4769dfb-c9e4-4f72-b3d6-1d18f4ac4ee6"
    oam_subnet_id: "191f7580-acf6-4c2b-8ec0-ba7d99b3bc4e"
    oam_network_cidr: "192.168.30.0/24"
    username: "vnf_user"
    api_key: "vnf_password"
    tenant_name: "vnfs"
    tenant_id: "47899782ed714295b1151681fdfd51f5"
    region: "RegionOne"
    keystone_url: "http://1.2.3.4:5000"
    flavour_medium: "m1.medium"
    service_tenant_name: "services"
  dmaap_topic: "AUTO"
  demo_artifacts_version: "1.1.0-SNAPSHOT"

The migration from the disparate configuration methodologies to Helm charts is tracked under the  OOM-460 - Getting issue details... STATUS  JIRA story. Within each of the projects a new configuration repository contains all of the project specific configuration artifacts.  As changes are made within the project, it's the responsibility of the project team to make appropriate changes to the configuration data.  The  OOM-10 - Getting issue details... STATUS  Epic tracks this work with the following status:

Monitoring

OOM provides two mechanims to monitor the real-time health of an ONAP deployment: a Consul GUI for a human operator or downstream monitoring systems and Kubernetes liveness probes that enable automatic healing of failed containers. The liveness probes are described in the following section.

OOM deploys a 3 instance Consul server cluster that provides a real-time health monitoring capability for all of the ONAP components.  For each of the ONAP components a Consul health check has been created, here is an example from the AAI model loader:

{
  "service": {
    "name": "A&AI Model Loader",
    "checks": [
      {
        "id": "model-loader-process",
        "name": "Model Loader Presence",
        "script": "/consul/config/scripts/model-loader-script.sh",
        "interval": "15s",
        "timeout": "1s"
      }
    ]
  }
}

To see the real-time health of a deployment go to: 

 http://<kubernetes IP>:30270/ui/

where a GUI much like the following will be found:

The  OOM-7 - Getting issue details... STATUS  JIRA Epic tracks this work with the following status:

Heal

Liveness probes are used by the Kubernetes manager to monitor the real-time health of the containers in an ONAP deployment.  If the liveness probe fails, Kubernetes will kill off the failed container and start a new container to replace it. The liveness probe have same parameters as the readiness probes described in the Deploy section. For example, to monitor the SDNC component has following liveness probe can be found in the SDNC DB deployment specification:

livenessProbe:
  exec:
    command: ["mysqladmin", "ping"]
  initialDelaySeconds: 30
  periodSeconds: 10
  timeoutSeconds: 5

Note that containers are inherently ephemeral so the healing action destroys failed containers and any state information within it.  To avoid a loss of state, a persistent volume should be used to store all data that needs to be persisted over the re-creation of a container.  Persistent volumes have been created for the database components of each of the projects and the same technique can be used for all persistent state information.

Clustering and Scaling

In-order to avoid the loss of an entire ONAP component by the failure of a single container, OOM enables the creation of clusters of pods hidden behind a load balancer in the form of a service.  As an example of what can be done, please refer to the  SDN-C Clustering on Kubernetes wiki page.

Upgrade 

OOM uses Helm K8S package manager to deploy ONAP components. Each component is arranged in a packaging format called a chart – a collection of files that describe a set of k8s resources. Helm allows for rolling upgrades of the ONAP component deployed. To upgrade a component Helm release you will need an updated Helm chart. The chart might have modified, deleted or added values, deployment yamls, and more.

To get the release name use:

helm ls

To easily upgrade the release use:

helm upgrade [RELEASE] [CHART]

To roll back to a previous release version use:

helm rollback [flags] [RELEASE] [REVISION]

for example, to upgrade the onap-mso helm release to the latest MSO container release v1.1.2:

• Edit mso valus.yaml which is part of the chart

Change “mso: nexus3.onap.org:10001/openecomp/mso:v1.1.1” to

“mso: nexus3.onap.org:10001/openecomp/mso:v1.1.2”

• From the chart location run:
  helm upgrade onap-mso ./
  
  

The previous mso pod will be terminated and a new mso pod with an updated mso container will be created.

Deletion