charts

ITOps Helm Charts

Helm chart repository for ITOps — a Kubernetes-native IT Operations Platform.

Usage

helm repo add itops https://charts.mlops.hu
helm repo update

Available Charts

Chart Version Description
itops/itops 1.16.1 Core + UI + bundled PostgreSQL
itops/itops-agent 1.4.1 K8s operator for service discovery and SLA reporting
itops/sla-portal 1.3.3 Standalone public status page (SQLite)

Quick Start

1. Deploy the platform

helm install itops itops/itops -n itops --create-namespace

Default login: admin / Password123!.

External exposure:

helm install itops itops/itops -n itops --create-namespace \
  --set ingress.hosts[0].host=api.yourdomain.com \
  --set ingress.tls[0].hosts[0]=api.yourdomain.com \
  --set uiIngress.hosts[0].host=app.yourdomain.com \
  --set uiIngress.tls[0].hosts[0]=app.yourdomain.com

2. Connect each cluster with the agent

helm install itops-agent itops/itops-agent \
  --set node.id="myorg/platform/prod/cluster1" \
  --set itops.url="https://api.yourdomain.com" \
  --set itops.apiKey.value="YOUR_OPERATOR_API_KEY" \
  -n itops --create-namespace

OPERATOR_API_KEY comes from the platform chart’s secretEnv.ITOPS_SECURITY_OPERATOR_API_KEY.

Multiple clusters are supported out of the box: each agent uses its own node.id prefix (4-level path org/platform/env/cluster) and pushes independently. The backend stores every service under externalId = nodeId/serviceName, so cross-cluster name collisions are impossible.

GitOps schema — one line minimum

A service’s it-ops.yaml needs exactly one required field: path. The 5-level identifier organization/platform/environment/cluster/service uniquely addresses the service across every cluster connected to the platform.

# The smallest valid it-ops.yaml
path: "myorg/itops/prod/cluster1/payment-api"

That’s it. The agent registers the service, the K8s workload health flows automatically. Every other field is opt-in — add a line for each feature you want.

Progressive disclosure

Add Effect
slaGroup: "..." SLA tracking turns on; criticality inherited from the group tier
backup: { expected: true, maxAgeDays: 1 } Backup tab entry with overdue alerts
type: api team: backend tags: [api] Presentation fields (icon, UI labels, filters)
dependencies.requires: [{ path: "..." }] Clickable dep chip in the UI

Dependency references are globally unique

Every dep ref uses the same path: identifier. No more name collisions between clusters:

path: "myorg/itops/prod/eu-west-1/payment-api"
slaGroup: "payment-system"

dependencies:
  requires:
    - path: "myorg/itops/prod/eu-west-1/payment-db"
      critical: true
  usedBy:
    - path: "myorg/itops/prod/us-east-1/payment-frontend"  # different cluster

Loud-over-silent push endpoints

The /api/v1/storage/report, /backup/report, /health/report webhooks now handle malformed payloads by padding to sentinels + returning warnings instead of dropping data. A push without nodeId lands under an unknown/unknown/unknown/unknown hierarchy branch with a red badge in the UI (so the misconfig is visible), and the response body echoes a warnings[] array explaining what was normalized. See monitoring docs for examples.

Documentation