GCP - create multiple VM using terraform - google-cloud-platform

Question

we need to create 2 VM in GCP using terraform with:
different startup scripts
we need the first and second VM will not be created at the same time.
we also need the IP of the first VM that was created and pass it in to the metadata script of the second VM.
How can we do it?
This is our working terraform file to create 1 VM:
provider "google" {
project = "project-name"
region = "europe-west3"
zone = "europe-west3-c"
}
resource "google_compute_instance" "vm_instance" {
name = "terra-instance-with-script-newvpc"
machine_type = "n1-standard-2"
zone = "europe-west3-c"
boot_disk {
initialize_params {
image ="centos-7"
}
}
metadata = {
startup-script = <<-EOF
hello world
EOF
}
network_interface {
subnetwork="test-subnet-frankfurt"
}
}
we tried to add another script and its failed

Answer 1

To ensure that the second GCE Instance is created after the first
one, you can use the depends_on argument (see the example code
below)
For the startup scripts, you can use a separate cloud-init script(s) and manage them using the Terraform variables, as
illustrated in the example code below.
For the IP address part, there are 2 possible use cases, depending on your needs:
Assign a Public Static IP: you can use the google_compute_address Terraform resource first to prepare an IP for the first instance, then call
this resource in the network_interface {access_config = { nat_ip = google_compute_address.static.first_instance_ip.address }} argument (see the
example below)
Then, in the init script of the second instance you can again call the Terraform resource google_compute_address.first_instance_ip.address and inject it into your script the way you want.
If you want to use an ephemeral (generated) IP instead, then you don’t need to use the google_compute_address at all, in your init script of the second instance, call the first instance Terraform resource with the nat ip attribute, e.g: google_compute_instance.firs_instance.network_interface.0.access_config.0.nat_ip and inject it in your script the way you want.
In the example code below, it assumes you will use a static IP address:
# Create the IP address
resource "google_compute_address" "static" {
name = "first_instance_ip_address"
}
# Source the init script and inject the IP Address variable
data "template_file" "init" {
template = "${file("./startup.sh")}" # call your startup script from it's path
vars = {
first_instance_ip = "${google_compute_address.static.address}"
# In case you use the generated IP:
# first_instance_ip = "${google_compute_compute_instance.first_instance.network_interface.0.access_config.0.nat_ip}"
}
}
# Create the 1st google compute instance
resource "google_compute_instance" "firs_instance" {
depends_on = [google_compute_address.static]
....
# Associated our public IP address to this instance
access_config = {
nat_ip = google_compute_address.static.address
}
}
...
# Create your second IP same as the first one
...
#Create the second google compute instance
resource "google_compute_instance" "second_instance" {
depends_on = [google_compute_instance.first_instance]
...
# Associated our public IP address to this instance
access_config = {
nat_ip = google_compute_address.second_static_ip.address # optional depends on your usecase
}
}
metadata_startup_script = "${data.template_file.startup_script.rendered}"
}
P.S: this is not a complete and production-ready code, it is a collection of tips to help you solve your issue.