Deploy your first service

This is the quick start with more meat on the bones. We'll deploy a real nginx service with health checks, an exposed port, and an environment variable.

1. Write the spec

nginx.yaml:

service:
  name: web
  namespace: default
  image: nginx:alpine
  scale: 2
 
  ports:
    - name: http
      port: 80
 
  env:
    NGINX_HOST: localhost
 
  resources:
    cpu:
      request: 100m
      limit: 500m
    memory:
      request: 64Mi
      limit: 256Mi
 
  health:
    liveness:
      type: http
      path: /
      port: 80
      initialDelaySeconds: 5
      intervalSeconds: 10
      timeoutSeconds: 2
      failureThreshold: 3

2. Lint before you cast

rune lint nginx.yaml

rune lint validates the schema and checks for common mistakes (typos in fields, invalid resource quantities, undefined references). Fix anything it flags before applying.

3. Apply

rune cast nginx.yaml

Output:

Service 'web' created in namespace 'default'.
Generation: 1
Waiting for rollout...
  ✓ web-instance-a4f9d2 Running
  ✓ web-instance-b7e3c1 Running
Rollout complete.

The CLI streams reconciliation progress until all instances are ready. Pass --detach to return immediately.

4. Verify

rune get services
rune get instances -n default
rune health web --checks
rune logs web --tail=20

rune health shows liveness probe results — useful when something looks wrong.

5. Iterate

Change scale: 2 to scale: 4, then re-cast:

rune cast nginx.yaml

The reconciler computes the diff, increments the generation, and rolls in two new instances.

For a quick scale-only operation, skip the YAML edit:

rune scale web 4

6. Inspect a single instance

rune get instances
rune get instance web-instance-a4f9d2 -o yaml

You'll see status conditions, restart count, started-at, and the runner-specific details.

7. Exec in for debugging

rune exec web
# or against a specific instance:
rune exec web-instance-a4f9d2 ls /etc/nginx

8. Tear down

rune delete web

Or stop without deleting (keeps spec, drops to 0 instances):

rune stop web
rune scale web 2   # bring it back later

What you've used so far

Where next

• Use secrets & configmaps — same nginx, but with real config.
• Health checks — designing probes that don't lie.
• Scale & restart — gradual rollouts.