one worker node per zone

[otrebor]

Dear everyone,

I am trying to configure terraform to deploy a worker node for each zone but i don't understand how to set the module to reach this goal. Can you please give me an hint on how to configure such behaviour?

[otrebor]

Just to give you a bit of context, I am trying to provision a cluster and install Openshift data foundation on it, via terraform

[toddgiguere]

@otrebor

The mapping of workers to subnets (zones) is accomplished through two variables:

First, the vpc_subnets, which is a map where each entry can contain one-or-more subnets from different zones.

Second input is the worker_pools map, where each pool (including "default") can have a property subnet_prefix which points to the map key from var.vpc_subnets, and also a workers_per_zone that specifies how many worker nodes should be put in each entry of var.vpc_subnets.

A simple multi-zone deployment (one per zone) would look something like this, with the subnet data coming from however you are deploying the VPC:

  vpc_subnets = {
    default = [
      {
        id         = ibm_is_subnet.subnet_zone_1.id
        cidr_block = ibm_is_subnet.subnet_zone_1.ipv4_cidr_block
        zone       = ibm_is_subnet.subnet_zone_1.zone
      },
      {
        id         = ibm_is_subnet.subnet_zone_2.id
        cidr_block = ibm_is_subnet.subnet_zone_2.ipv4_cidr_block
        zone       = ibm_is_subnet.subnet_zone_2.zone
      },
      {
        id         = ibm_is_subnet.subnet_zone_3.id
        cidr_block = ibm_is_subnet.subnet_zone_3.ipv4_cidr_block
        zone       = ibm_is_subnet.subnet_zone_3.zone
      }
    ]
  }

  worker_pools = [
    {
      subnet_prefix    = "default" # this value points to the map key for subnets above
      pool_name        = "default" # ibm_container_vpc_cluster automatically names default pool "default" 
      machine_type     = "bx2.4x16"
      workers_per_zone = 1 
      operating_system = "RHCOS"
    }
  ]

This should result in a cluster with one pool (called "default"), with one worker node in each of the 3 zones defined in vpc_subnets.

A very basic example of this can be seen in our basic example with a single zone and 2 workers.

You can use subnets and pool inputs to create an even more complex solution if you wish to have multiple pools with different machine or storage types. You can see an example of this in our advanced example.

Let me know if I need to explain further. I don't know if there is a need for Data Foundation to have a mixture of different worker pools? If so then I can explain how that advanced example works a little more.

[toddgiguere]

@otrebor I'm closing this issue as its been a while with no further questions. If you still have issues you can contact me and I'll re-open.