Migrate from Harness

Concept map

Harness	Slipway	Notes
Account / Organization / Project	Organisation / Project
Pipeline, Stage, Step	Pipeline, Stage, Job, Step
CI stage (Build)	A job (or several) running steps
CD stage (Deploy)	A deploy job with environment:	Strategy runs in the deploy engine
Service	Service	Artifact source maps to the service’s artifact spec
Environment	Environment	Environment type (prod / non-prod) maps to kind
Infrastructure Definition	Infrastructure	Kubernetes, ECS, Lambda, Cloud Run, Azure, VM, Terraform
Delegate	Runner	Dials out, like the delegate; pools and labels replace delegate selectors
Delegate selector	runs_on labels
Connector	Integration
Secrets Manager	Secrets with external references; BYOK keyring
Rolling / Blue-Green / Canary deployment	rolling / bluegreen / canary
Continuous Verification	Canary verification	Provider queries, baselines and thresholds
Approval step (Harness / Jira / ServiceNow)	Approval, change-ticket gate
Step Group, Looping strategy (matrix, repeat)	matrix
Conditional execution	if: with CEL
Failure strategy	retry, allow_failure, auto_rollback
Input sets, runtime inputs	manual / api inputs
Pipeline / Step templates	Org templates, extends	Versioned and immutable
Triggers (webhook, scheduled, artifact)	on: triggers; release trigger for new releases
OPA policies	CEL policy gates (external OPA planned)
Service variables, environment overrides	Variables scoped by service and environment
Feature Flags	Flag integrations (LaunchDarkly, Unleash, Flagsmith)	Slipway does not implement flags itself
Cloud Cost Management, Chaos, STO	Not in scope	Integrate dedicated tools

Worked example

A Harness pipeline with a Build stage, then a Deploy stage to a Kubernetes prod environment using canary with Prometheus verification and a Jira approval becomes:

version: 1
name: api
on: { push: { branches: [main] } }
jobs:
  - name: build
    runs_on: [linux, docker]
    outputs: { version: "${{ steps.v.outputs.version }}" }
    steps:
      - uses: slipway/checkout@v1
      - id: v
        run: echo "version=1.4.${{ git.short_sha }}" >> "$SLIPWAY_OUTPUT"
      - uses: slipway/docker-build@v1
      - uses: slipway/release-create@v1
  - name: deploy-prod
    needs: [build]
    environment: { name: production, service: api }
    steps:
      - uses: slipway/deploy@v1
        with: { release: "${{ needs.build.outputs.version }}", strategy: canary }

and in the app:

• Production environment: required_approvals: 1, change_ticket_required: true with the Jira integration (Business and above) in place of the Jira approval step.
• Canary strategy config: steps 5 / 25 / 100 and a verify block using the Prometheus provider. See Canary verification. Translate Harness’s metric definitions into error_rate and relative metrics.

Steps

1. Install runners in the networks your delegates cover; recreate delegate selectors as labels.
2. Recreate connectors as integrations; prefer OIDC and external references over stored credentials.
3. Model services and environments and attach infrastructure.
4. Convert pipelines, starting with CI stages.
5. Port verification by recreating metric queries and thresholds, then dry-run against recent deployments.
6. Cut over environment by environment.