This repo is intended to assist with peformance testing for pycytominer.
This repo contains terraform and ansible configuration for setting up a single VM with a networking layer.
You'll need to install Terraform and Ansible prior to using the scripts in this repo. You'll also need a GNU environment, including bash or a bash-compatible shell.
While it's not required, you'll probably want to install the Google Cloud CLI
for your platform. Follow the CLI installation
instructions and be sure to log
in via gcloud auth login once it's installed.
In order to provsion the resources specified in this repo to GCP, you'll need a
service key with Compute Engine create permissions. You should obtain a service
key with the necessary permissions, name it sa.json, and put it in
./deployment/.secrets.
Run ./deployment/provision.sh to first run terraform to provision the
infrastructure, then ansible to deploy software to it. To perform an unattended
fresh reinstall, you can pass the -d (for "destroy") and -x (automatically
accept) flags to the script, i.e.:
./deployment/provision.sh -x -d
After a few minutes, you should have a provisioned VM. The state of this VM and
any other created resources (e.g., network configuration) are stored in an
untracked file called ./deployment/terraform.tfstate; don't check this file in
or otherwise share it as it can contain sensitive information.
The Ansible playbook at ./deployment/ansible/setup.yml will also be run
against the inventory of VMs created by Terraform, but initially it does nothing
but check that the hosts are online and make them echo a greeting. Feel free to
modify it to your needs.
Assuming you have the Google Cloud (aka "gcloud") CLI installed, you can SSH into your machine with the following command:
gcloud compute ssh --project=cuhealthai-sandbox --zone=us-central1-a test-sample-vm
When you're done using your resources and want to tear down the infrastructure,
the following command will destroy the resources and not recreate them (-na,
"no apply", skips the Terraform "apply" step that creates resources):
./deployment/provision.sh -x -d -na
First off, you'll find project-level variables specified in
./deployment/terraform/config.tg. Feel free to modify them in that file, or
override their values through Terraform's many methods for specifying module
variable
values.
Most of the infrastructure specification lives in
./deployment/terraform/machines.tf; modify the virtual_machines variable to
your needs.
You'll find network configuration in ./deployment/terraform/network.tf; if you
need to open additional ports on your VM, you'd add a tag description there,
then add the tag to your VM's list of tags. See "fw_http_tag" for an example, in
that case of opening port 80 to any client.
Finally, ./deployment/terraform/outputs.tf allows you to specify information
you'd like to gather via Terraform that you'd want to persist to another module.
The current configuration produces a dictionary of GCP instance name to its
external IP address as the output. The values outputted via outputs.tf are
available as host-level variables in Ansible; for example, in the current config
"{{ vars.outputs.instance_ip_addr['test-vm-sample-vm'] }}" would resolve to the
external IP of the test-vm-sample-vm instance.