Implement Kubernetes recommended labels for all chart components

[djfarrelly]

Summary

• Implement proper semantic labeling following Kubernetes best practices
• Add component-specific label helpers for inngest server, PostgreSQL, and Redis
• Fix app.kubernetes.io/name to identify actual applications, not the chart name

Description

This PR updates the Helm chart to follow Kubernetes recommended labels conventions,
ensuring proper semantic labeling across all components.

Changes

Label Schema:

Component	app.kubernetes.io/name	app.kubernetes.io/component	app.kubernetes.io/part-of
Inngest Server	inngest	server	inngest
PostgreSQL	postgresql	database	inngest
Redis	redis	queue	inngest

New Helper Functions in _helpers.tpl:

• inngest.server.labels / inngest.server.selectorLabels
• inngest.postgresql.labels / inngest.postgresql.selectorLabels
• inngest.redis.labels / inngest.redis.selectorLabels

Files Modified: 18 files across all template resources

Why This Matters

Previously, all components used the same app.kubernetes.io/name: inngest label, which:

• Made it difficult to query/filter resources by actual application
• Didn't follow Kubernetes semantics where name should identify the application
• Required manual app.kubernetes.io/component additions that were inconsistent

Now each component has proper labels that enable:

• Filtering all PostgreSQL resources: app.kubernetes.io/name=postgresql
• Filtering all Inngest platform resources: app.kubernetes.io/part-of=inngest
• Understanding component roles: app.kubernetes.io/component=database

⚠️ Breaking Change

This is a breaking change for existing deployments.

Changing app.kubernetes.io/name in selector labels causes Kubernetes to reject updates to existing Deployments because selector fields are immutable after creation.

Upgrade Path:

Users upgrading from previous versions must delete and recreate their deployments:

# Option 1: Delete specific deployments
kubectl delete deployment inngest inngest-postgresql inngest-redis -n inngest
helm upgrade <release-name> .

# Option 2: Full reinstall
helm uninstall <release-name>
helm install <release-name> .

[djfarrelly]

Ensure that traffic is routed to only the server pods, not any other pod like Redis.