Change worker managed instance group to span zones in region

* Change Google Cloud module to require the `region` variable
* Workers are created in random zones within the given region
* Tolerate Google Cloud zone failures or capacity issues
* If workers are preempted (if enabled), replacement instances can
be drawn from any zone in the region, which should avoid scheduling
issues that were possible before if a single zone aggressively
preempts instances (presumably due to Google Cloud capacity)

CHANGES.md

@@ -4,6 +4,11 @@ Notable changes between versions.
 
 ## Latest
 
+#### Google Cloud
+
+* Add required variable `region`
+* Change worker managed instance group to automatically span zones in a region
+
 ## v1.8.2
 
 * Kubernetes v1.8.2

README.md

@@ -46,6 +46,7 @@ module "google-cloud-yavin" {
   source = "git::https://github.com/poseidon/typhoon//google-cloud/container-linux/kubernetes"
 
   # Google Cloud
+  region        = "us-central1"
   zone          = "us-central1-c"
   dns_zone      = "example.com"
   dns_zone_name = "example-zone"

docs/google-cloud.md

@@ -77,6 +77,7 @@ module "google-cloud-yavin" {
   source = "git::https://github.com/poseidon/typhoon//google-cloud/container-linux/kubernetes"
 
   # Google Cloud
+  region        = "us-central1"
   zone          = "us-central1-c"
   dns_zone      = "example.com"
   dns_zone_name = "example-zone"
@@ -196,6 +197,7 @@ Learn about [version pinning](concepts.md#versioning), maintenance, and [addons]
 | Name | Description | Example |
 |:-----|:------------|:--------|
 | cluster_name | Unique cluster name (prepended to dns_zone) | "yavin" |
+| region | Google Cloud region | "us-central1" |
 | zone | Google Cloud zone | "us-central1-f" |
 | dns_zone | Google Cloud DNS zone | "google-cloud.example.com" |
 | dns_zone_name | Google Cloud DNS zone name | "example-zone" |

docs/index.md

@@ -46,6 +46,7 @@ module "google-cloud-yavin" {
   source = "git::https://github.com/poseidon/typhoon//google-cloud/container-linux/kubernetes"
 
   # Google Cloud
+  region        = "us-central1"
   zone          = "us-central1-c"
   dns_zone      = "example.com"
   dns_zone_name = "example-zone"

google-cloud/container-linux/kubernetes/cluster.tf

@@ -29,8 +29,8 @@ module "workers" {
 
   # GCE
   network      = "${google_compute_network.network.name}"
+  region       = "${var.region}"
   count        = "${var.worker_count}"
-  zone         = "${var.zone}"
   machine_type = "${var.machine_type}"
   os_image     = "${var.os_image}"
   preemptible  = "${var.worker_preemptible}"

google-cloud/container-linux/kubernetes/variables.tf

@@ -3,9 +3,14 @@ variable "cluster_name" {
   description = "Cluster name"
 }
 
+variable "region" {
+  type        = "string"
+  description = "Google Cloud Region (e.g. us-central1, see `gcloud compute regions list`)"
+}
+
 variable "zone" {
   type        = "string"
-  description = "Google Cloud zone (e.g. us-central1-f, see `gcloud compute zones list`)"
+  description = "Google Cloud Zone (e.g. us-central1-f, see `gcloud compute zones list`)"
 }
 
 variable "dns_zone" {

google-cloud/container-linux/kubernetes/workers/variables.tf

@@ -20,9 +20,9 @@ variable "count" {
   description = "Number of worker compute instances the instance group should manage"
 }
 
-variable "zone" {
+variable "region" {
   type        = "string"
-  description = "Google zone that compute instances in the group should be created in (e.g. gcloud compute zones list)"
+  description = "Google Cloud region to create a regional managed group of workers (e.g. us-central1, see `gcloud compute regions list`)."
 }
 
 variable "machine_type" {

google-cloud/container-linux/kubernetes/workers/workers.tf

@@ -1,16 +1,17 @@
-# Managed Instance Group
+# Regional managed instance group maintains a homogeneous set of workers that
-resource "google_compute_instance_group_manager" "workers" {
+# span the zones in the region.
+resource "google_compute_region_instance_group_manager" "workers" {
   name        = "${var.cluster_name}-worker-group"
   description = "Compute instance group of ${var.cluster_name} workers"
 
-  # Instance name prefix for instances in the group
+  # instance name prefix for instances in the group
   base_instance_name = "${var.cluster_name}-worker"
   instance_template  = "${google_compute_instance_template.worker.self_link}"
-  update_strategy    = "RESTART"
+  region             = "${var.region}"
-  zone               = "${var.zone}"
+
   target_size = "${var.count}"
 
-  # Target pool instances in the group should be added into
+  # target pool to which instances in the group should be added
   target_pools = [
     "${google_compute_target_pool.workers.self_link}",
   ]