This repository contains Terraform templates for paving the necessary infrastructure to deploy Tanzu application Platform (PKS and TAS) to a single foundation. The templates support AWS, vSphere, Azure, and GCP.
As of v3.0.0
of this repository the following are the supported versions of the Terraform CLI
and providers:
v1.0.11
or higherProvider versions | Provider | version |
---|---|---|
GCP | 4.1.0 | |
AWS | 3.65.0 | |
Azure | 2.85.0 |
In each IaaS directory, there is a terraform.tfvars.example
you can copy
and modify with your configuration choices and credentials.
terraform init
terraform plan -var-file terraform.tfvars
terraform apply -var-file terraform.tfvars
terraform output stable_config_output
terraform destroy -var-file terraform.tfvars
The terraform templates are namespaced for the resources that consume them.
In each IAAS, the prefix opsmanager-
, pks-
, and pas-
are on the file names.
There are cases that some resources aren't required in a foundation.
For example, just deploying PKS and not PAS.
To remove PAS resources, just rm pas-*.tf
the file from the directory.
Please note that the opsmanager-*.tf
files cannot be removed.
Every foundation requires an Ops Manager.
These templates support deploying Tanzu Application Service (TAS) and Pivotal Container Service (PKS) to the same foundation.
The templates do not create an Ops Manager VM but do create the necessary infrastructure for the VM (security groups, keys, etc).
These templates demonstrate a modest production deployment in three (3) AZs on each IaaS.
These templates contain extremely minimal interdependence or cleverness, to facilitate incorporating these templates into your own automation easily.
The semantics of the versioning of paving's releases are based on the contents
of terraform output stable_config_(opsmanager|pas|pks)
. stable_config
should always represent
the minimum necessary to install Pivotal Platform. Any other output may be
added or removed without a change in version. However, MAJOR.MINOR.PATCH should
change according to the following:
In our current configuration, we are using the Ops Manager VM as the
jumpbox. The Ops Manager VM is deployed in the public subnet with a
configuration (var.ops_manager_allowed_ips
) to restrict it by IP. If you want to use a
jumpbox instead, you may deploy ops manager in the management subnet.