Kubernetes Wait for Deployment

Waits for Kubernetes workloads to become ready with automatic Argo Rollout delegation.

Features

• ⏳ Universal workload support - Deployments, StatefulSets, DaemonSets, and Rollouts
• 🔄 Automatic Rollout delegation - Seamlessly handles Argo Rollouts via wait-for-rollout
• 📊 Batch processing - Wait for multiple workloads from kustomize-inspect output
• 🎯 Smart namespace handling - Automatic namespace resolution
• 📝 Status reporting - Detailed workload status after waiting

Usage

- name: Wait for deployments
  uses: skyhook-io/wait-for-deployment@v1
  with:
    namespace: production
    workloads_json: '[{"kind":"Deployment","name":"api"},{"kind":"StatefulSet","name":"db"}]'
    timeout: 5m

Inputs

Input	Description	Required	Default
namespace	Default namespace for workloads	✅	-
workloads_json	JSON array of workloads: [{kind,name,namespace?}]	✅	-
timeout	Wait timeout (e.g., 300s, 5m)	❌	300s
show_status	Show kubectl status after waiting	❌	true
verify_only	Pass-through for Rollout verification	❌	false
wait_for_deployment_id	Wait for deployment-id annotation before checking rollout status	❌	false
expected_deployment_id	Expected deployment-id annotation value	❌	github.run_id
annotation_timeout	Timeout for annotation wait (e.g., 300s, 5m)	❌	300s

Supported Workload Types

• Deployment - Standard Kubernetes deployments
• StatefulSet - Stateful applications
• DaemonSet - Node-level workloads
• Rollout - Argo Rollouts (delegated to wait-for-rollout@v1)

Examples

Basic Deployment Wait

- name: Deploy and wait
  run: kubectl apply -f manifests/

- name: Wait for deployment
  uses: skyhook-io/wait-for-deployment@v1
  with:
    namespace: production
    workloads_json: '[{"kind":"Deployment","name":"frontend"}]'

Multiple Workloads

- name: Wait for all services
  uses: skyhook-io/wait-for-deployment@v1
  with:
    namespace: default
    workloads_json: |
      [
        {"kind":"Deployment","name":"api"},
        {"kind":"Deployment","name":"worker"},
        {"kind":"StatefulSet","name":"database"}
      ]
    timeout: 10m

With Kustomize Inspect

- name: Inspect manifests
  id: inspect
  uses: skyhook-io/kustomize-inspect@v1
  with:
    overlay_path: overlays/production

- name: Wait for workloads
  uses: skyhook-io/wait-for-deployment@v1
  with:
    namespace: production
    workloads_json: ${{ steps.inspect.outputs.workloads }}

Mixed Standard and Rollout Workloads

- name: Wait for mixed workloads
  uses: skyhook-io/wait-for-deployment@v1
  with:
    namespace: production
    workloads_json: |
      [
        {"kind":"Deployment","name":"frontend"},
        {"kind":"Rollout","name":"api"},
        {"kind":"StatefulSet","name":"cache"}
      ]

With Custom Namespaces

- name: Wait across namespaces
  uses: skyhook-io/wait-for-deployment@v1
  with:
    namespace: default  # Fallback for workloads without namespace
    workloads_json: |
      [
        {"kind":"Deployment","name":"api","namespace":"backend"},
        {"kind":"Deployment","name":"web","namespace":"frontend"},
        {"kind":"StatefulSet","name":"db"}
      ]

ArgoCD Async Deployments

When deploying via ArgoCD (or other async GitOps tools), the workload may not be updated immediately after triggering a sync. Use wait_for_deployment_id to wait for the correct version before checking rollout status:

- name: Trigger ArgoCD sync
  run: argocd app sync my-app --async

- name: Wait for deployment
  uses: skyhook-io/wait-for-deployment@v1
  with:
    namespace: production
    workloads_json: ${{ steps.inspect.outputs.workloads }}
    wait_for_deployment_id: 'true'  # Waits for deployment-id annotation
    # expected_deployment_id defaults to github.run_id

This requires your manifests to include a deployment-id annotation (typically set via kustomize):

apiVersion: apps/v1
kind: Deployment
metadata:
  annotations:
    deployment-id: "12345678"  # Set to github.run_id
spec:
  template:
    metadata:
      annotations:
        deployment-id: "12345678"  # Also on pod template

How It Works

1. Parses workload list - Processes the JSON array of workloads
2. Separates by type - Identifies standard K8s workloads vs Argo Rollouts
3. Waits for annotation (if enabled) - Polls until deployment-id annotation matches expected value
4. Delegates Rollouts - Automatically uses wait-for-rollout@v1 for Rollout resources
5. Waits for standard workloads - Uses kubectl rollout status for Deployments, StatefulSets, DaemonSets
6. Reports status - Shows final status for all workloads

Rollout Delegation

When a Rollout is detected in the workload list:

• Automatically delegates to skyhook-io/wait-for-rollout@v1
• Passes through timeout and verify_only parameters
• Supports only one Rollout per action invocation (use matrix for multiple)

Error Handling

The action will fail if:

• Any Deployment fails to become ready
• Timeout is exceeded
• Multiple Rollouts are provided (use matrix strategy instead)
• kubectl connection fails

Note: StatefulSet and DaemonSet failures are non-blocking by default.

Complete Deployment Flow

jobs:
  deploy:
    steps:
      - name: Checkout
        uses: actions/checkout@v4
      
      - name: Setup kubectl
        uses: skyhook-io/cloud-login@v1
        with:
          provider: aws
          region: us-east-1
          cluster: production
      
      - name: Build manifests
        id: build
        run: |
          kustomize build overlays/production > manifests.yaml
          kubectl apply -f manifests.yaml
      
      - name: Get workloads
        id: inspect
        uses: skyhook-io/kustomize-inspect@v1
        with:
          overlay_path: overlays/production
      
      - name: Wait for all workloads
        uses: skyhook-io/wait-for-deployment@v1
        with:
          namespace: production
          workloads_json: ${{ steps.inspect.outputs.workloads }}
          timeout: 10m
      
      - name: Run tests
        if: success()
        run: ./scripts/smoke-test.sh

Prerequisites

• Kubernetes cluster access (configured kubectl)
• For Rollouts: Argo Rollouts installed in cluster
• Appropriate RBAC permissions

Notes

• Requires kubectl context to be configured
• Use with cloud-login action for authentication
• For multiple Rollouts, use matrix strategy or sequential calls
• Timeout applies to each workload wait operation