feat: Enable Surge Upgrades by specifying max_surge and max_unavailable (Beta) (#394)

BREAKING CHANGE: beta clusters now have surge upgrades turned on by default. This behavior can be tuned using the max_surge and max_unavailable inputs.

Author: Dev25

autogen/main/README.md

@@ -189,6 +189,8 @@ The node_pools variable takes the following parameters:
 | max_count | Maximum number of nodes in the NodePool. Must be >= min_count | 100 | Optional |
 {% if beta_cluster %}
 | max_pods_per_node | The maximum number of pods per node in this cluster | null | Optional |
+| max_surge | The number of additional nodes that can be added to the node pool during an upgrade. Increasing max_surge raises the number of nodes that can be upgraded simultaneously. Can be set to 0 or greater. | 1 | Optional |
+| max_unavailable | The number of nodes that can be simultaneously unavailable during an upgrade. Increasing max_unavailable raises the number of nodes that can be upgraded in parallel. Can be set to 0 or greater. | 0 | Optional |
 {% endif %}
 | min_count | Minimum number of nodes in the NodePool. Must be >=0 and <= max_count. Should be used when autoscaling is true | 1 | Optional |
 | name | The name of the node pool |  | Required |

autogen/main/cluster.tf.tmpl

@@ -382,6 +382,13 @@ resource "google_container_node_pool" "pools" {
     auto_upgrade = lookup(each.value, "auto_upgrade", local.default_auto_upgrade)
   }
 
+  {% if beta_cluster %}
+  upgrade_settings {
+    max_surge = lookup(each.value, "max_surge", 1)
+    max_unavailable = lookup(each.value, "max_unavailable", 0)
+  }
+  {% endif %}
+
   node_config {
     image_type   = lookup(each.value, "image_type", "COS")
     machine_type = lookup(each.value, "machine_type", "n1-standard-2")

cluster.tf

@@ -158,6 +158,7 @@ resource "google_container_node_pool" "pools" {
     auto_upgrade = lookup(each.value, "auto_upgrade", local.default_auto_upgrade)
   }
 
+
   node_config {
     image_type   = lookup(each.value, "image_type", "COS")
     machine_type = lookup(each.value, "machine_type", "n1-standard-2")

modules/beta-private-cluster-update-variant/README.md

@@ -276,6 +276,8 @@ The node_pools variable takes the following parameters:
 | machine_type | The name of a Google Compute Engine machine type | n1-standard-2 | Optional |
 | max_count | Maximum number of nodes in the NodePool. Must be >= min_count | 100 | Optional |
 | max_pods_per_node | The maximum number of pods per node in this cluster | null | Optional |
+| max_surge | The number of additional nodes that can be added to the node pool during an upgrade. Increasing max_surge raises the number of nodes that can be upgraded simultaneously. Can be set to 0 or greater. | 1 | Optional |
+| max_unavailable | The number of nodes that can be simultaneously unavailable during an upgrade. Increasing max_unavailable raises the number of nodes that can be upgraded in parallel. Can be set to 0 or greater. | 0 | Optional |
 | min_count | Minimum number of nodes in the NodePool. Must be >=0 and <= max_count. Should be used when autoscaling is true | 1 | Optional |
 | name | The name of the node pool |  | Required |
 | node_count | The number of nodes in the nodepool when autoscaling is false. Otherwise defaults to 1. Only valid for non-autoscaling clusers |  | Required |

modules/beta-private-cluster-update-variant/cluster.tf

@@ -346,6 +346,11 @@ resource "google_container_node_pool" "pools" {
     auto_upgrade = lookup(each.value, "auto_upgrade", local.default_auto_upgrade)
   }
 
+  upgrade_settings {
+    max_surge       = lookup(each.value, "max_surge", 1)
+    max_unavailable = lookup(each.value, "max_unavailable", 0)
+  }
+
   node_config {
     image_type   = lookup(each.value, "image_type", "COS")
     machine_type = lookup(each.value, "machine_type", "n1-standard-2")

modules/beta-private-cluster/README.md

@@ -254,6 +254,8 @@ The node_pools variable takes the following parameters:
 | machine_type | The name of a Google Compute Engine machine type | n1-standard-2 | Optional |
 | max_count | Maximum number of nodes in the NodePool. Must be >= min_count | 100 | Optional |
 | max_pods_per_node | The maximum number of pods per node in this cluster | null | Optional |
+| max_surge | The number of additional nodes that can be added to the node pool during an upgrade. Increasing max_surge raises the number of nodes that can be upgraded simultaneously. Can be set to 0 or greater. | 1 | Optional |
+| max_unavailable | The number of nodes that can be simultaneously unavailable during an upgrade. Increasing max_unavailable raises the number of nodes that can be upgraded in parallel. Can be set to 0 or greater. | 0 | Optional |
 | min_count | Minimum number of nodes in the NodePool. Must be >=0 and <= max_count. Should be used when autoscaling is true | 1 | Optional |
 | name | The name of the node pool |  | Required |
 | node_count | The number of nodes in the nodepool when autoscaling is false. Otherwise defaults to 1. Only valid for non-autoscaling clusers |  | Required |

modules/beta-private-cluster/cluster.tf

@@ -274,6 +274,11 @@ resource "google_container_node_pool" "pools" {
     auto_upgrade = lookup(each.value, "auto_upgrade", local.default_auto_upgrade)
   }
 
+  upgrade_settings {
+    max_surge       = lookup(each.value, "max_surge", 1)
+    max_unavailable = lookup(each.value, "max_unavailable", 0)
+  }
+
   node_config {
     image_type   = lookup(each.value, "image_type", "COS")
     machine_type = lookup(each.value, "machine_type", "n1-standard-2")

modules/beta-public-cluster/README.md

@@ -232,6 +232,8 @@ The node_pools variable takes the following parameters:
 | machine_type | The name of a Google Compute Engine machine type | n1-standard-2 | Optional |
 | max_count | Maximum number of nodes in the NodePool. Must be >= min_count | 100 | Optional |
 | max_pods_per_node | The maximum number of pods per node in this cluster | null | Optional |
+| max_surge | The number of additional nodes that can be added to the node pool during an upgrade. Increasing max_surge raises the number of nodes that can be upgraded simultaneously. Can be set to 0 or greater. | 1 | Optional |
+| max_unavailable | The number of nodes that can be simultaneously unavailable during an upgrade. Increasing max_unavailable raises the number of nodes that can be upgraded in parallel. Can be set to 0 or greater. | 0 | Optional |
 | min_count | Minimum number of nodes in the NodePool. Must be >=0 and <= max_count. Should be used when autoscaling is true | 1 | Optional |
 | name | The name of the node pool |  | Required |
 | node_count | The number of nodes in the nodepool when autoscaling is false. Otherwise defaults to 1. Only valid for non-autoscaling clusers |  | Required |

modules/beta-public-cluster/cluster.tf

@@ -261,6 +261,11 @@ resource "google_container_node_pool" "pools" {
     auto_upgrade = lookup(each.value, "auto_upgrade", local.default_auto_upgrade)
   }
 
+  upgrade_settings {
+    max_surge       = lookup(each.value, "max_surge", 1)
+    max_unavailable = lookup(each.value, "max_unavailable", 0)
+  }
+
   node_config {
     image_type   = lookup(each.value, "image_type", "COS")
     machine_type = lookup(each.value, "machine_type", "n1-standard-2")

modules/private-cluster-update-variant/cluster.tf

@@ -243,6 +243,7 @@ resource "google_container_node_pool" "pools" {
     auto_upgrade = lookup(each.value, "auto_upgrade", local.default_auto_upgrade)
   }
 
+
   node_config {
     image_type   = lookup(each.value, "image_type", "COS")
     machine_type = lookup(each.value, "machine_type", "n1-standard-2")