[doc] 3.3.1 Environment Description - load balancer message "wrong"?

[Martin-Weiss]

3.3.1 Environment Description

https://susedoc.github.io/doc-caasp/master/single-html/caasp-deployment/#_environment_description

---

This seems to be outdated:

This will be added in future versions. You must provide your own load balancing solution that directs access to the master nodes.

-> we have the nginx and haproxy load balancer examples in this guide - so maybe just link it?

[r0ckarong]

This box is related to the VMWare deployment and we do not provide any information on how to set up load balancer in a VMWare environment (yet). From what I recall we rely on the VMWare administrator to know how to set up load balancing in their environment. I will change the wording slightly to reflrect this box is directly talking about VMWare.

[Martin-Weiss]

This box is related to the VMWare deployment and we do not provide any information on how to set up load balancer in a VMWare environment (yet). From what I recall we rely on the VMWare administrator to know how to set up load balancing in their environment. I will change the wording slightly to reflrect this box is directly talking about VMWare.

See https://susedoc.github.io/doc-caasp/master/single-html/caasp-deployment/#loadbalancer - this is not "environment specific" at all. And even in the terraform for VMware we deliver "segments" of the load balancer configuration for VMware.. ;-).

[r0ckarong]

@thehejik @kkaempf Are both nginx and haproxy tested on VMWare? If so can I just assume they can be deployed like for any other platform?

[kkaempf]

@fgerling - can you answer this ?

[thehejik]

It wasn't tested for quite some time.

The info in doc should be suitable for any platform if you configure it manually.

We are not delivering haproxy nor nginx code in terraform states to customers, although skuba master branch has the haproxy lb tf code available for vmware and kvm/libvirt but for development purposes only (we are removing it by skuba patch in IBS).

[r0ckarong]

It wasn't tested for quite some time.

The info in doc should be suitable for any platform if you configure it manually.

We are not delivering haproxy nor nginx code in terraform states to customers, although skuba master branch has the haproxy lb tf code available for vmware and kvm/libvirt but for development purposes only (we are removing it by skuba patch in IBS).

What is the correct procedure to deploy a load balancer for a VMWare environment? Deploy the same template you use for the rest of the cluster nodes and deploy LB software on it and configure? And then proceed with the actual cluster setup?

[Martin-Weiss]

Either setup a single VM with SLES and install haproxy package or setup a SLE-HA cluster in VMs.

Then configure haproxy based on the documented haproxy.conf adjusted to the names of your masters/workers.

Create a DNS entry for "api" pointing to the load balancers VIP and then deploy the cluster using skuba.

P.S. in the meantime we might not need to document / recommend the nginx "workaround" anymore - as haproxy supported is available via SLE-HA (if customer wants it supported) or haproxy is also available "unsupported" via the packagehub channel...