enos: use on-demand targets (#21459) (#21463)

Add an updated `target_ec2_instances` module that is capable of dynamically splitting target instances over subnet/az's that are compatible with the AMI architecture and the associated instance type for the architecture. Use the `target_ec2_instances` module where necessary. Ensure that `raft` storage scenarios don't provision unnecessary infrastructure with a new `target_ec2_shim` module. After a lot of trial, the state of Ec2 spot instance capacity, their associated APIs, and current support for different fleet types in AWS Terraform provider, have proven to make using spot instances for scenario targets too unreliable. The current state of each method: * `target_ec2_fleet`: unusable due to the fact that the `instant` type does not guarantee fulfillment of either `spot` or `on-demand` instance request types. The module does support both `on-demand` and `spot` request types and is capable of bidding across a maximum of four availability zones, which makes it an attractive choice if the `instant` type would always fulfill requests. Perhaps a `request` type with `wait_for_fulfillment` option like `aws_spot_fleet_request` would make it more viable for future consideration. * `target_ec2_spot_fleet`: more reliable if bidding for target instances that have capacity in the chosen zone. Issues in the AWS provider prevent us from bidding across multiple zones succesfully. Over the last 2-3 months target capacity for the instance types we'd prefer to use has dropped dramatically and the price is near-or-at on-demand. The volatility for nearly no cost savings means we should put this option on the shelf for now. * `target_ec2_instances`: the most reliable method we've got. It is now capable of automatically determing which subnets and availability zones to provision targets in and has been updated to be usable for both Vault and Consul targets. By default we use the cheapest medium instance types that we've found are reliable to test vault. * Update .gitignore * enos/modules/create_vpc: create a subnet for every availability zone * enos/modules/target_ec2_fleet: bid across the maximum of four availability zones for targets * enos/modules/target_ec2_spot_fleet: attempt to make the spot fleet bid across more availability zones for targets * enos/modules/target_ec2_instances: create module to use ec2:RunInstances for scenario targets * enos/modules/target_ec2_shim: create shim module to satisfy the target module interface * enos/scenarios: use target_ec2_shim for backend targets on raft storage scenarios * enos/modules/az_finder: remove unsed module Signed-off-by: Ryan Cragun <me@ryan.ec> Co-authored-by: Ryan Cragun <me@ryan.ec>
hashicorp · Jun 26, 2023 · 84d2bb1 · 84d2bb1
1 parent 39eb1d6
commit 84d2bb1
Show file tree

Hide file tree

Showing 17 changed files with 511 additions and 89 deletions.
diff --git a/.gitignore b/.gitignore
@@ -60,18 +60,13 @@ Vagrantfile
 !enos/**/*.hcl
 
 # Enos
-enos/.enos
-enos/enos-local.vars.hcl
-enos/support
-# Enos local Terraform files
-enos/.terraform/*
-enos/.terraform.lock.hcl
-enos/*.tfstate
-enos/*.tfstate.*
-enos/**/.terraform/*
-enos/**/.terraform.lock.hcl
-enos/**/*.tfstate
-enos/**/*.tfstate.*
+.enos
+enos-local.vars.hcl
+enos/**/support
+enos/**/kubeconfig
+.terraform
+.terraform.lock.hcl
+.tfstate.*
 
 .DS_Store
 .idea

diff --git a/enos/enos-modules.hcl b/enos/enos-modules.hcl
@@ -58,27 +58,40 @@ module "shutdown_multiple_nodes" {
   source = "./modules/shutdown_multiple_nodes"
 }
 
+# create target instances using ec2:CreateFleet
 module "target_ec2_fleet" {
   source = "./modules/target_ec2_fleet"
 
-  capacity_type    = "on-demand" // or "spot", use on-demand until we can stabilize spot fleets
-  common_tags      = var.tags
-  instance_mem_min = 4096
-  instance_cpu_min = 2
-  max_price        = "0.1432" // On-demand cost for RHEL amd64 on t3.medium in us-east
-  project_name     = var.project_name
-  ssh_keypair      = var.aws_ssh_keypair_name
+  common_tags  = var.tags
+  project_name = var.project_name
+  ssh_keypair  = var.aws_ssh_keypair_name
 }
 
+# create target instances using ec2:RunInstances
+module "target_ec2_instances" {
+  source = "./modules/target_ec2_instances"
+
+  common_tags  = var.tags
+  project_name = var.project_name
+  ssh_keypair  = var.aws_ssh_keypair_name
+}
+
+# don't create instances but satisfy the module interface
+module "target_ec2_shim" {
+  source = "./modules/target_ec2_shim"
+
+  common_tags  = var.tags
+  project_name = var.project_name
+  ssh_keypair  = var.aws_ssh_keypair_name
+}
+
+# create target instances using ec2:RequestSpotFleet
 module "target_ec2_spot_fleet" {
   source = "./modules/target_ec2_spot_fleet"
 
-  common_tags      = var.tags
-  instance_mem_min = 4096
-  instance_cpu_min = 2
-  max_price        = "0.1432" // On-demand cost for RHEL amd64 on t3.medium in us-east
-  project_name     = var.project_name
-  ssh_keypair      = var.aws_ssh_keypair_name
+  common_tags  = var.tags
+  project_name = var.project_name
+  ssh_keypair  = var.aws_ssh_keypair_name
 }
 
 module "vault_agent" {

diff --git a/enos/enos-scenario-agent.hcl b/enos/enos-scenario-agent.hcl
@@ -90,7 +90,7 @@ scenario "agent" {
   }
 
   step "create_vault_cluster_targets" {
-    module     = module.target_ec2_spot_fleet
+    module     = module.target_ec2_instances
     depends_on = [step.create_vpc]
 
     providers = {

diff --git a/enos/enos-scenario-autopilot.hcl b/enos/enos-scenario-autopilot.hcl
@@ -101,7 +101,7 @@ scenario "autopilot" {
   }
 
   step "create_vault_cluster_targets" {
-    module     = module.target_ec2_spot_fleet
+    module     = module.target_ec2_instances
     depends_on = [step.create_vpc]
 
     providers = {
@@ -194,7 +194,7 @@ scenario "autopilot" {
   }
 
   step "create_vault_cluster_upgrade_targets" {
-    module     = module.target_ec2_spot_fleet
+    module     = module.target_ec2_instances
     depends_on = [step.create_vpc]
 
     providers = {

diff --git a/enos/enos-scenario-replication.hcl b/enos/enos-scenario-replication.hcl
@@ -110,7 +110,7 @@ scenario "replication" {
 
   # Create all of our instances for both primary and secondary clusters
   step "create_primary_cluster_targets" {
-    module = module.target_ec2_spot_fleet
+    module = module.target_ec2_instances
     depends_on = [
       step.create_vpc,
     ]
@@ -129,7 +129,7 @@ scenario "replication" {
   }
 
   step "create_primary_cluster_backend_targets" {
-    module = module.target_ec2_spot_fleet
+    module = matrix.primary_backend == "consul" ? module.target_ec2_instances : module.target_ec2_shim
     depends_on = [
       step.create_vpc,
     ]
@@ -139,7 +139,7 @@ scenario "replication" {
     }
 
     variables {
-      ami_id                = step.ec2_info.ami_ids["amd64"]["ubuntu"]["22.04"]
+      ami_id                = step.ec2_info.ami_ids["arm64"]["ubuntu"]["22.04"]
       awskms_unseal_key_arn = step.create_vpc.kms_key_arn
       cluster_tag_key       = local.backend_tag_key
       common_tags           = local.tags
@@ -148,7 +148,7 @@ scenario "replication" {
   }
 
   step "create_primary_cluster_additional_targets" {
-    module = module.target_ec2_spot_fleet
+    module = module.target_ec2_instances
     depends_on = [
       step.create_vpc,
       step.create_primary_cluster_targets,
@@ -169,7 +169,7 @@ scenario "replication" {
   }
 
   step "create_secondary_cluster_targets" {
-    module     = module.target_ec2_spot_fleet
+    module     = module.target_ec2_instances
     depends_on = [step.create_vpc]
 
     providers = {
@@ -186,15 +186,15 @@ scenario "replication" {
   }
 
   step "create_secondary_cluster_backend_targets" {
-    module     = module.target_ec2_spot_fleet
+    module     = matrix.secondary_backend == "consul" ? module.target_ec2_instances : module.target_ec2_shim
     depends_on = [step.create_vpc]
 
     providers = {
       enos = provider.enos.ubuntu
     }
 
     variables {
-      ami_id                = step.ec2_info.ami_ids["amd64"]["ubuntu"]["22.04"]
+      ami_id                = step.ec2_info.ami_ids["arm64"]["ubuntu"]["22.04"]
       awskms_unseal_key_arn = step.create_vpc.kms_key_arn
       cluster_tag_key       = local.backend_tag_key
       common_tags           = local.tags

diff --git a/enos/enos-scenario-smoke.hcl b/enos/enos-scenario-smoke.hcl
@@ -111,7 +111,7 @@ scenario "smoke" {
   }
 
   step "create_vault_cluster_targets" {
-    module     = module.target_ec2_spot_fleet
+    module     = module.target_ec2_instances
     depends_on = [step.create_vpc]
 
     providers = {
@@ -128,15 +128,15 @@ scenario "smoke" {
   }
 
   step "create_vault_cluster_backend_targets" {
-    module     = module.target_ec2_spot_fleet
+    module     = matrix.backend == "consul" ? module.target_ec2_instances : module.target_ec2_shim
     depends_on = [step.create_vpc]
 
     providers = {
       enos = provider.enos.ubuntu
     }
 
     variables {
-      ami_id                = step.ec2_info.ami_ids["amd64"]["ubuntu"]["22.04"]
+      ami_id                = step.ec2_info.ami_ids["arm64"]["ubuntu"]["22.04"]
       awskms_unseal_key_arn = step.create_vpc.kms_key_arn
       cluster_tag_key       = local.backend_tag_key
       common_tags           = local.tags

diff --git a/enos/enos-scenario-ui.hcl b/enos/enos-scenario-ui.hcl
@@ -78,7 +78,7 @@ scenario "ui" {
   }
 
   step "create_vault_cluster_targets" {
-    module     = module.target_ec2_spot_fleet
+    module     = module.target_ec2_instances
     depends_on = [step.create_vpc]
 
     providers = {
@@ -95,15 +95,15 @@ scenario "ui" {
   }
 
   step "create_vault_cluster_backend_targets" {
-    module     = module.target_ec2_spot_fleet
+    module     = matrix.backend == "consul" ? module.target_ec2_instances : module.target_ec2_shim
     depends_on = [step.create_vpc]
 
     providers = {
       enos = provider.enos.ubuntu
     }
 
     variables {
-      ami_id                = step.ec2_info.ami_ids["amd64"]["ubuntu"]["22.04"]
+      ami_id                = step.ec2_info.ami_ids["arm64"]["ubuntu"]["22.04"]
       awskms_unseal_key_arn = step.create_vpc.kms_key_arn
       cluster_tag_key       = local.backend_tag_key
       common_tags           = local.tags

diff --git a/enos/enos-scenario-upgrade.hcl b/enos/enos-scenario-upgrade.hcl
@@ -106,7 +106,7 @@ scenario "upgrade" {
   }
 
   step "create_vault_cluster_targets" {
-    module     = module.target_ec2_spot_fleet
+    module     = module.target_ec2_instances
     depends_on = [step.create_vpc]
 
     providers = {
@@ -123,15 +123,15 @@ scenario "upgrade" {
   }
 
   step "create_vault_cluster_backend_targets" {
-    module     = module.target_ec2_spot_fleet
+    module     = matrix.backend == "consul" ? module.target_ec2_instances : module.target_ec2_shim
     depends_on = [step.create_vpc]
 
     providers = {
       enos = provider.enos.ubuntu
     }
 
     variables {
-      ami_id                = step.ec2_info.ami_ids["amd64"]["ubuntu"]["22.04"]
+      ami_id                = step.ec2_info.ami_ids["arm64"]["ubuntu"]["22.04"]
       awskms_unseal_key_arn = step.create_vpc.kms_key_arn
       cluster_tag_key       = local.backend_tag_key
       common_tags           = local.tags

diff --git a/enos/modules/create_vpc/main.tf b/enos/modules/create_vpc/main.tf
@@ -1,4 +1,11 @@
-data "aws_region" "current" {}
+data "aws_availability_zones" "available" {
+  state = "available"
+
+  filter {
+    name   = "zone-name"
+    values = ["*"]
+  }
+}
 
 resource "random_string" "cluster_id" {
   length  = 8
@@ -34,14 +41,16 @@ resource "aws_vpc" "vpc" {
 }
 
 resource "aws_subnet" "subnet" {
+  count                   = length(data.aws_availability_zones.available.names)
   vpc_id                  = aws_vpc.vpc.id
-  cidr_block              = var.cidr
+  cidr_block              = cidrsubnet(var.cidr, 8, count.index)
+  availability_zone       = data.aws_availability_zones.available.names[count.index]
   map_public_ip_on_launch = true
 
   tags = merge(
     var.common_tags,
     {
-      "Name" = "${var.name}-subnet"
+      "Name" = "${var.name}-subnet-${data.aws_availability_zones.available.names[count.index]}"
     },
   )
 }

diff --git a/enos/modules/create_vpc/outputs.tf b/enos/modules/create_vpc/outputs.tf
@@ -1,8 +1,3 @@
-output "aws_region" {
-  description = "AWS Region for resources"
-  value       = data.aws_region.current.name
-}
-
 output "vpc_id" {
   description = "Created VPC ID"
   value       = aws_vpc.vpc.id

diff --git a/enos/modules/target_ec2_fleet/main.tf b/enos/modules/target_ec2_fleet/main.tf
@@ -76,15 +76,23 @@ resource "random_string" "unique_id" {
   special = false
 }
 
+// ec2:CreateFleet only allows up to 4 InstanceRequirements overrides so we can only ever request
+// a fleet across 4 or fewer subnets if we want to bid with InstanceRequirements instead of
+// weighted instance types.
+resource "random_shuffle" "subnets" {
+  input        = data.aws_subnets.vpc.ids
+  result_count = 4
+}
+
 locals {
-  spot_allocation_strategy      = "price-capacity-optimized"
+  spot_allocation_strategy      = "lowestPrice"
   on_demand_allocation_strategy = "lowestPrice"
   instances                     = toset([for idx in range(var.instance_count) : tostring(idx)])
   cluster_name                  = coalesce(var.cluster_name, random_string.random_cluster_name.result)
   name_prefix                   = "${var.project_name}-${local.cluster_name}-${random_string.unique_id.result}"
   fleet_tag                     = "${local.name_prefix}-spot-fleet-target"
   fleet_tags = {
-    Name                     = "${local.name_prefix}-target"
+    Name                     = "${local.name_prefix}-${var.cluster_tag_key}-target"
     "${var.cluster_tag_key}" = local.cluster_name
     Fleet                    = local.fleet_tag
   }
@@ -218,6 +226,20 @@ resource "aws_launch_template" "target" {
     name = aws_iam_instance_profile.target.name
   }
 
+  instance_requirements {
+    burstable_performance = "included"
+
+    memory_mib {
+      min = var.instance_mem_min
+      max = var.instance_mem_max
+    }
+
+    vcpu_count {
+      min = var.instance_cpu_min
+      max = var.instance_cpu_max
+    }
+  }
+
   network_interfaces {
     associate_public_ip_address = true
     delete_on_termination       = true
@@ -251,7 +273,9 @@ resource "aws_launch_template" "target" {
 # Unless we see capacity issues or instances being shut down then we ought to
 # stick with that strategy.
 resource "aws_ec2_fleet" "targets" {
-  terminate_instances = true // termiante instances when we "delete" the fleet
+  replace_unhealthy_instances         = false
+  terminate_instances                 = true // terminate instances when we "delete" the fleet
+  terminate_instances_with_expiration = false
   tags = merge(
     var.common_tags,
     local.fleet_tags,
@@ -264,22 +288,11 @@ resource "aws_ec2_fleet" "targets" {
       version            = aws_launch_template.target.latest_version
     }
 
-    override {
-      max_price = var.max_price
-      subnet_id = data.aws_subnets.vpc.ids[0]
-
-      instance_requirements {
-        burstable_performance = "included"
-
-        memory_mib {
-          min = var.instance_mem_min
-          max = var.instance_mem_max
-        }
+    dynamic "override" {
+      for_each = random_shuffle.subnets.result
 
-        vcpu_count {
-          min = var.instance_cpu_min
-          max = var.instance_cpu_max
-        }
+      content {
+        subnet_id = override.value
       }
     }
   }

diff --git a/enos/modules/target_ec2_fleet/variables.tf b/enos/modules/target_ec2_fleet/variables.tf
@@ -25,7 +25,7 @@ variable "common_tags" {
   description = "Common tags for cloud resources"
   type        = map(string)
   default = {
-    Project = "Vault"
+    Project = "vault-ci"
   }
 }