/nsx-container-plugin-operator

Kubernetes Operator for the NSX Container Plugin (NCP)

Primary LanguageGoOtherNOASSERTION

NSX Container Plugin Operator

License

Overview

An operator for leveraging NSX as the default container networking solution for an Kubernetes/Openshift cluster. The operator will be deployed in the early phases of Openshift cluster deployment or after the kubectl is ready in Kubernetes cluster, and it will take care of deploying NSX integration components, and precisely:

  • The NSX container plugin (NCP) deployment
  • The nsx-ncp-bootstrap daemonset
  • The nsx-node-agent daemonset

The nsx-container-plugin operator monitors a dedicated ConfigMap, applies changes to NCP and nsx-node-agent configuration, and creates/restarts the relevant pods so that the relevant configuration changes are picked up.

The nsx-container-plugin operator also monitors the nsx-node-agent status and updates the network status on relevant nodes.

In addition, the nsx-container-plugin operator is able to monitor nodes ensuring the corresponding NSX logical port is enabled as a container host logical port.

For Openshift 4 clusters, the nsx-container-plugin operator especially monitors the network.config.openshift.io CR to update the container network CIDRs used by NCP.

Try it out

Preparing the operator image

Pull the packed image for docker:

docker pull vmware/nsx-container-plugin-operator:latest

For containerd:

ctr image pull docker.io/vmware/nsx-container-plugin-operator:latest

Building the nsx-container-plugin operator is very simple. From the project root directory simply type the following command, which based on docker build tool.

make all

At the moment the nsx-container-plugin operator only works on native Kubernetes or Openshift 4 environments

Installing

Kubernetes

Edit the operator yaml files in deploy/kubernetes then apply them.

Openshift

Installing a cluster with user-provisioned infrastructure
  1. Preparing install-config.yaml Generate install-config.yaml by using openshift-install command.
$ openshift-install --dir=$MY_CLUSTER create install-config

Edit $MY_CLUSTER/install-config.yaml to update networking section. Change networkType to ncp(case insensitive). Set container network CIDRs clusterNetwork in $MY_CLUSTER/install-config.yaml.

  1. Creating manifest files:
$ openshift-install --dir=$MY_CLUSTER create manifests

If one cluster node has multiple VirtualNetworkInterfaces, the operator cannot detect which interface should be enabled as the containers' parent interface, so the user should edit deploy/openshift4/operator.nsx.vmware.com_v1_ncpinstall_cr.yaml to set addNodeTag: false and manually tag the target node port by scope=ncp/node_name, tag=<node_name> and scope=ncp/node_name, tag=<cluster_name> on NSX-T.

Put operator yaml files from deploy/openshift4/ to $MY_CLUSTER/manifests, edit configmap.yaml about operator configurations, add the operator image and NCP image in operator.yaml.

  1. Generating ignition configuration files:
$ openshift-install --dir=$MY_CLUSTER create ignition-configs

This bootstrap ignition file will be added to the terraform tfvars. Then use terraform to install Openshift 4 cluster on vSphere.

Installing a cluster with installer-provisioned infrastructure
  1. Prepare install-config.yaml This step is similar to UPI installation. An example of install-config.yaml:
apiVersion: v1
baseDomain: openshift.test
compute:
- architecture: amd64
  hyperthreading: Enabled
  name: worker
  platform: {}
  replicas: 3
controlPlane:
  architecture: amd64
  hyperthreading: Enabled
  name: master
  platform: {}
  replicas: 3
metadata:
  creationTimestamp: null
  name: ipi
networking:
  networkType: ncp
  clusterNetwork:
  - cidr: 10.0.0.0/14
    hostPrefix: 24
  machineCIDR: 192.168.10.0/24
  serviceNetwork:
  - 172.8.0.0/16
platform:
  vsphere:
    apiVIP: 192.168.10.11
    cluster: cluster
    datacenter: dc
    defaultDatastore: vsanDatastore
    ingressVIP: 192.168.10.12
    network: openshift-segment
    password: pass
    username: user
    vCenter: my-vc.local
publish: External
pullSecret: 'xxx'
sshKey: 'ssh-rsa xxx'

You can validate your DNS configuration before installing OpenShift Container Platform on IPI. A sample DNS zone database as follow:

$TTL    604800

$ORIGIN openshift.test.
@       IN      SOA     dns1.openshift.test. root.openshift.test. (
                              2         ; Serial
                         604800         ; Refresh
                          86400         ; Retry
                        2419200         ; Expire
                         604800 )       ; Negative Cache TTL
; main domain name servers
@       IN      NS      localhost.
@       IN      A       127.0.0.1
@       IN      AAAA    ::1
        IN      NS      dns1.openshift.test.

; recors for name servers above
dns1    IN      A       10.92.204.129

; sub-domain definitions
$ORIGIN ipi.openshift.test.
api IN A 192.168.10.11
apps IN A 192.168.10.12

; sub-domain definitions
$ORIGIN apps.ipi.openshift.test.
* IN A 192.168.10.12
  1. Preparing manifest files:

Put operator yaml files from deploy/openshift4/ to $MY_CLUSTER/manifests, edit configmap.yaml about operator configurations, add the operator image and NCP image in operator.yaml.

  1. Creating cluster
$ openshift-install create cluster --dir=$MY_CLUSTER

The installation log locates in $MY_CLUSTER/.openshift_install.log. If the deployment ends in timeout or failure, you can check the environment according to the log, then Re-run Installer to continue to get the installation log:

$ openshift-install wait-for install-complete

Upgrade

For upgrading, all yaml files in deploy/${platform}/ should be involved, especially to check the image and NCP_IMAGE in `deploy/${platform}/operator.yaml

Documentation

Cluster network config (Openshift specific)

Cluster network config is initially set in install-config.yaml, user could apply network.config.openshift.io CRD to update clusterNetwork in manifests/cluster-network-02-config.yml. Example configurations

apiVersion: config.openshift.io/v1
kind: Network
metadata:
  name: cluster
spec:
  clusterNetwork:
  - cidr: 10.10.0.0/14
  networkType: ncp

Operator ConfigMap

Operator ConfigMap nsx-ncp-operator-config is used to provide NCP configurations. As for now we only support NSX Policy API, single Tier topology on Openshift 4, single or two Tiers topology on native Kubernetes.

Kubernetes

Some fields are mandatory including cluster, nsx_api_managers, container_ip_blocks, tier0_gateway(for single T1 case), top_tier_router (for single T0 case), external_ip_pools(for SNAT mode).. If any of above options is not provided in the operator ConfigMap, the operator will fail to reconcile configurations, error message swill be added in ncpinstall nsx-ncp Degraded conditions

OpenShift

The operator sets policy_nsxapi as True, single_tier_topology as True. In the ConfigMap, some fields are mandatory including cluster, nsx_api_managers, tier0_gateway(for single T1 case), top_tier_router(for single T0 case), external_ip_pools(for SNAT mode). If any of above options is not provided in the operator ConfigMap, the operator will fail to reconcile configurations, error messages will be added in clusteroperator nsx-ncp Degraded conditions.

NCP Image

User needs to set NCP image as an environment parameter NCP_IMAGE in deploy/${platform}/operator.yaml.

Unsafe changes

  • (Openshift specific) If CIDRs in clusterNetwork are already applied, it is unsafe to remove them. NSX NCP operator won't fail when it detects some existing network CIDRs are deleted, but the removal may cause unexpected issues.
  • NSX NCP operator uses tags to mark the container host logical ports, deleting these tags from NSX manager will cause network realization failure on corresponding nodes.

Contributing

We welcome community contributions to the NSX Container plugin Operator!

Before you start working with nsx-container-plugin-operator, you should sign our contributor license agreement (CLA).

If you wish to contribute code and you have not signed our CLA, our bot will update the issue when you open a Pull Request. For more detailed information, refer to CONTRIBUTING.md.

For any questions about the CLA process, please refer to our FAQ.

License

This repository is available under the Apache 2.0 license.