Ops-controller-manager Overview

ops-controller-manager is a Kubernetes Operator that provides three core objects: Host, Cluster, and Task.

Objects in ops-controller-manager

Host: Describes a host machine, including hostname, IP address, SSH username, password, private key, etc.
Cluster: Describes a cluster, including details such as cluster name, number of hosts, number of pods, load, CPU, memory, etc.
Task: Describes a task, which can be one-time or scheduled (cron) tasks.

Installation

Prerequisites

Kubernetes 1.19+
Helm 3.0+
NATS server (optional but recommended for event streaming)

Install Helm

If you haven't installed Helm yet, you can install it using the following command:

curl https://raw.githubusercontent.com/helm/helm/master/scripts/get-helm-3 | bash

Add Helm Repository

Add the ops Helm repository to your Helm configuration:

helm repo add ops https://www.chenshaowen.com/ops/charts
helm repo update

Install ops-controller-manager

Basic Installation:

To install the ops-controller-manager using Helm with default values:

helm install myops ops/ops --version 2.0.0 --namespace ops-system --create-namespace

Installation with Custom Values:

You can customize the installation using --set flags:

helm install myops ops/ops --version 2.0.0 \
  --namespace ops-system \
  --create-namespace \
  --set controller.env.eventEndpoint="http://app:password@nats-headless.ops-system.svc:4222" \
  --set replicaCount=2 \
  --set resources.limits.memory=4096Mi

Installation with Values File:

Create a custom values file for more complex configurations:

# my-values.yaml
replicaCount: 2

controller:
  env:
    activeNamespace: "ops-system"
    eventEndpoint: "http://app:password@nats-headless.ops-system.svc:4222"
    defaultRuntimeImage: "registry.cn-beijing.aliyuncs.com/opshub/ubuntu:22.04"

server:
  env:
    eventEndpoint: "http://app:password@nats-headless.ops-system.svc:4222"

resources:
  limits:
    cpu: 2000m
    memory: 4096Mi
  requests:
    cpu: 1000m
    memory: 2048Mi

prometheus:
  enabled: true

autoscaling:
  enabled: true
  minReplicas: 2
  maxReplicas: 10
  targetCPUUtilizationPercentage: 80

Then install with the values file:

helm install myops ops/ops --version 2.0.0 \
  --namespace ops-system \
  --create-namespace \
  -f my-values.yaml

Verify Installation

After the installation, check the status of the pods to ensure everything is running:

# Check pods
kubectl get pods -n ops-system

# Check services
kubectl get svc -n ops-system

# Check deployments
kubectl get deployments -n ops-system

Upgrade Installation

To upgrade an existing installation:

helm upgrade myops ops/ops --version 2.0.0 --namespace ops-system

Or with custom values:

helm upgrade myops ops/ops --version 2.0.0 \
  --namespace ops-system \
  -f my-values.yaml

Uninstall ops-controller-manager

To uninstall the ops-controller-manager:

helm uninstall myops --namespace ops-system

Configuration

Namespace Configuration

By default, ops-controller-manager will only process CRD resources within the ops-system namespace.

To change this behavior, you can modify the controller.env.activeNamespace value in the Helm values. If left empty, it will process resources from all namespaces.

Using --set:

helm install myops ops/ops --version 2.0.0 \
  --namespace ops-system \
  --create-namespace \
  --set controller.env.activeNamespace=""

Using values file:

controller:
  env:
    activeNamespace: ""  # Empty means all namespaces

Event Configuration

Configure NATS event endpoint for both controller and server:

controller:
  env:
    eventCluster: "default"
    eventEndpoint: "http://app:password@nats-headless.ops-system.svc:4222"

server:
  env:
    eventCluster: "default"
    eventEndpoint: "http://app:password@nats-headless.ops-system.svc:4222"

Resource Configuration

Configure resource limits and requests:

resources:
  limits:
    cpu: 2000m
    memory: 4096Mi
  requests:
    cpu: 1000m
    memory: 2048Mi

You can also configure separate resources for server:

server:
  resources:
    limits:
      cpu: 1000m
      memory: 2048Mi
    requests:
      cpu: 500m
      memory: 1024Mi

Monitoring Configuration

Enable Prometheus monitoring (creates ServiceMonitor resources):

prometheus:
  enabled: true

Autoscaling Configuration

Enable Horizontal Pod Autoscaler:

autoscaling:
  enabled: true
  minReplicas: 2
  maxReplicas: 10
  targetCPUUtilizationPercentage: 80
  targetMemoryUtilizationPercentage: 80

Troubleshooting

Check Pod Logs

# Controller logs
kubectl logs -n ops-system deployment/myops-ops

# Server logs
kubectl logs -n ops-system deployment/myops-ops-server

Check Metrics Endpoints

# Controller metrics
kubectl port-forward -n ops-system svc/myops-ops-controller-metrics 8080:8080
curl http://localhost:8080/metrics

# Server metrics
kubectl port-forward -n ops-system svc/myops-ops-server 8080:80
curl http://localhost:8080/metrics

Check ServiceMonitors

kubectl get servicemonitor -n ops-system

Opscontroller