Title: [Helm Chart 15.6.0 or newer] Can we scale zammad-scheduler + zammad-websocket, does zammad-railsserver support auto-scaling, and can zammad-nginx autoscale?
Infos:
- Used Zammad version: 6.5.1
- Used Zammad installation type: Helm Chart - 15.6.0
- Operating system: Kubernetes - v1.31.4
- Browser + version: Google Chrome - Version 145.0.7632.160 (Official Build) (arm64)
Expected behavior:
- Confirm whether the Helm deployment supports horizontal scaling / autoscaling (HPA) for these components:
zammad-scheduler: can/should it run with replicas > 1, or must it be a singleton to avoid duplicate jobs?zammad-websocket: can it be scaled to multiple replicas behind a Service/Ingress, and are sticky sessions / special LB settings required?zammad-railsserver: can it run with multiple replicas and be managed by HPA, and are there constraints (sessions, cache/Redis, DB connections, shared storage) to be aware of?zammad-nginx: can it be run with multiple replicas and autoscaled via HPA safely (stateless), and are there any recommended resource/connection settings when scaling?
Actual behavior:
- We haven’t changed replicas yet because we’re not sure what is supported.
- Our goal is to handle higher traffic and want to know the correct scaling approach for these components in Kubernetes.
Steps to reproduce the behavior:
- Deploy Zammad via Helm Chart 15.6.0 on Kubernetes.
- Consider increasing replicas / enabling HPA for:
zammad-schedulerzammad-websocketzammad-railsserverzammad-nginx
- Before applying changes in production, we want guidance on what is safe/supported and any required configuration.
