aws · gmarciani · Mar 13, 2025 · Mar 12, 2025 · Mar 12, 2025 · Mar 12, 2025
@@ -53,9 +53,9 @@ def self.configure(chef_run)
   for_all_oses do |platform, version|
     context "on #{platform}#{version}" do
       cached(:prerequisites) do
-        if %(redhat rocky).include?(platform)
+        if %(redhat rocky).include?(platform) || platform == 'amazon' && version == '2023'
           %w(environment-modules libibverbs-utils librdmacm-utils rdma-core-devel)
-        elsif platform == 'amazon'
+        elsif platform == 'amazon' && version == '2'
           %w(environment-modules libibverbs-utils librdmacm-utils)
         else
           "environment-modules"

@@ -27,19 +27,28 @@ def self.reload(chef_run)
       end
       cached(:node) { chef_run.node }
       cached(:network_service_name) do
-        {
-          'amazon' => 'network',
-          'centos' => 'network',
-          'redhat' => 'NetworkManager',
-          'rocky' => 'NetworkManager',
-          'ubuntu' => 'systemd-resolved',
-        }[platform]
+        if platform == 'amazon' && version == '2' || platform == 'centos'
+          'network'
+        elsif platform == 'amazon' && version == '2023'
+          'systemd-networkd'
+        elsif platform == 'ubuntu'
+          'systemd-resolved'
+        elsif %(redhat rocky).include?(platform)
+          'NetworkManager'
+        else
+          raise "Cannot determine network_service_name: unrecognized platform #{platform}"
+        end
       end
 
       it "restarts network service" do
         is_expected.to restart_network_service('restart')
+        network_services_to_restart = if platform == 'amazon' && version == '2023'
+                                        [network_service_name, 'systemd-resolved']
+                                      else
+                                        [network_service_name]
+                                      end
 
-        is_expected.to write_log("Restarting '#{network_service_name}' service, platform #{platform} '#{node['platform_version']}'")
+        is_expected.to write_log("Restarting '#{network_services_to_restart.join(' ')}' service, platform #{platform} '#{node['platform_version']}'")
 
         is_expected.to restart_service(network_service_name)
           .with(ignore_failure: true)
@@ -56,13 +65,17 @@ def self.reload(chef_run)
         ConvergeNetworkService.reload(runner)
       end
       cached(:network_service_name) do
-        {
-          'amazon' => 'network',
-          'centos' => 'network',
-          'redhat' => 'NetworkManager',
-          'rocky' => 'NetworkManager',
-          'ubuntu' => 'systemd-resolved',
-        }[platform]
+        if platform == 'amazon' && version == '2' || platform == 'centos'
+          'network'
+        elsif platform == 'amazon' && version == '2023'
+          'systemd-networkd'
+        elsif platform == 'ubuntu'
+          'systemd-resolved'
+        elsif %(redhat rocky).include?(platform)
+          'NetworkManager'
+        else
+          raise "Cannot determine network_service_name: unrecognized platform #{platform}"
+        end
       end
 
       it 'reloads network_service' do

@@ -11,7 +11,7 @@ def wait_for_block_dev(_path)
     context "on #{platform}#{version}" do
       cached(:venv_path) { 'venv' }
       cached(:raid_superblock_version) do
-        %(redhat rocky).include?(platform) || "#{platform}#{version}" == 'ubuntu20.04' || "#{platform}#{version}" == 'ubuntu22.04' ? '1.2' : '0.90'
+        %(redhat rocky ubuntu).include?(platform) || "#{platform}#{version}" == 'amazon2023' ? '1.2' : '0.90'
       end
       cached(:chef_run) do
         runner = runner(

@@ -26,14 +26,19 @@ def self.setup(chef_run)
         when 'ubuntu'
           "Ubuntu-#{version}"
         when 'amazon'
-          "AmazonLinux-2"
+          if version == '2'
+            "AmazonLinux-2"
+          elsif version == '2023'
+            'RHEL-9'
+          end
         else
           "RHEL-#{version}"
         end
       end
 
       cached(:gcc_major_minor_version) do
-        if platform == 'ubuntu' && version == '22.04' || version == '9'
+        case "#{platform}#{version}"
+        when 'amazon2023', 'ubuntu24.04', 'ubuntu22.04', 'redhat9', 'rocky9'
           '11.3'
         else
           '9.3'

@@ -42,7 +42,8 @@ def self.nothing(chef_run)
           allow_any_instance_of(Object).to receive(:arm_instance?).and_return(true)
         end
 
-        if platform == 'ubuntu' && version.to_i == 20
+        case "#{platform}#{version}"
+        when "amazon2023", "ubuntu20.04"
           it "is false" do
             expect(resource.dcv_supported?).to eq(false)
           end
@@ -58,9 +59,10 @@ def self.nothing(chef_run)
       end
 
       context 'when not on arm' do
-        it "is true" do
+        is_supported = !("#{platform}#{version}" == 'amazon2023')
+        it "is #{is_supported}" do
           allow_any_instance_of(Object).to receive(:arm_instance?).and_return(false)
-          expect(resource.dcv_supported?).to eq(true)
+          expect(resource.dcv_supported?).to eq(is_supported)
         end
       end
     end
@@ -142,8 +144,8 @@ def self.nothing(chef_run)
           expect(resource.xdcv).to eq("nice-xdcv_#{xdcv_version}_#{dcv_pkg_arch}.#{base_os}.deb")
           expect(resource.dcv_web_viewer).to eq("nice-dcv-web-viewer_#{dcv_webviewer_version}_#{dcv_pkg_arch}.#{base_os}.deb")
           expect(resource.dcv_gl).to eq("/nice-dcv-gl_#{dcv_gl_version}_#{dcv_pkg_arch}.#{base_os}.deb")
-        else
-          dcv_platform_version = platform == "amazon" ? "7" : version.to_i
+        elsif "#{platform}#{version}" != 'amazon2023'
+          dcv_platform_version = "#{platform}#{version}" == "amazon2" ? "7" : version.to_i
           dcv_platform_version_pkg = platform == "amazon" ? "amzn2" : "el" + version
           expect(resource.dcv_package).to eq("nice-dcv-#{dcv_version}-#{dcv_platform_version_pkg}-#{dcv_url_arch}")
           expect(resource.dcv_server).to eq("nice-dcv-server-#{dcv_server_version}.el#{dcv_platform_version}.#{dcv_url_arch}.rpm")
@@ -479,14 +481,16 @@ def self.nothing(chef_run)
                                                       .with_code(/apt -y purge ifupdown/)
                                                       .with_code(%r{wget https://d1uj6qtbmh3dt5.cloudfront.net/NICE-GPG-KEY})
           when 'amazon'
-            is_expected.to install_package(alinux_prereq_packages).with_retries(10).with_retry_delay(5)
-            is_expected.to create_file('Setup Gnome standard').with(
-              content: "PREFERRED=/usr/bin/gnome-session",
-              owner: "root",
-              group: "root",
-              mode: "0755",
-              path: "/etc/sysconfig/desktop"
-            )
+            if version == '2'
+              is_expected.to install_package(alinux_prereq_packages).with_retries(10).with_retry_delay(5)
+              is_expected.to create_file('Setup Gnome standard').with(
+                content: "PREFERRED=/usr/bin/gnome-session",
+                owner: "root",
+                group: "root",
+                mode: "0755",
+                path: "/etc/sysconfig/desktop"
+              )
+            end
           else
             is_expected.to run_execute('Install gnome desktop').with_command('yum -y install @gnome').with_retries(3).with_retry_delay(5)
             is_expected.to install_package('xorg-x11-server-Xorg').with_retries(3).with_retry_delay(5)

@@ -201,6 +201,12 @@ def self.configure(chef_run)
         case platform
         when 'ubuntu'
           %w(build-essential devscripts debhelper check libsubunit-dev fakeroot pkg-config dkms)
+        when 'amazon'
+          if version == '2023'
+            %w(dkms rpm-build make check check-devel)
+          else
+            %w(dkms rpm-build make check check-devel subunit subunit-devel)
+          end
         else
           %w(dkms rpm-build make check check-devel subunit subunit-devel)
         end
@@ -209,13 +215,15 @@ def self.configure(chef_run)
       cached(:gdrcopy_platform) do
         platforms = {
           'amazon2' => 'amzn-2',
+          'amazon2023' => 'amzn-2023',
           'centos7' => 'el7',
           'rhel8' => 'el8',
           'rocky8' => 'el8',
           'rhel9' => 'el9',
           'rocky9' => 'el9',
           'ubuntu20.04' => 'Ubuntu20_04',
           'ubuntu22.04' => 'Ubuntu22_04',
+          'ubuntu24.04' => 'Ubuntu24_04',
         }
         platforms["#{platform}#{version}"]
       end

@@ -59,7 +59,7 @@ def self.setup(chef_run)
           end
         end
 
-        if platform == 'amazon'
+        if platform == 'amazon' && version == '2'
           it 'installs extra packages' do
             is_expected.to install_alinux_extras_topic('R3.4')
           end

@@ -38,6 +38,8 @@
           "/usr/share/modules/init/.modulespath"
         when 'redhat', 'rocky'
           '/etc/environment-modules/modulespath'
+        when 'amazon'
+          version == '2' ? '/usr/share/Modules/init/.modulespath' : '/etc/environment-modules/modulespath'
         else
           "/usr/share/Modules/init/.modulespath"
         end

@@ -89,7 +89,7 @@ def self.setup(chef_run, nvidia_enabled: nil)
           chef_run.find_resource('nvidia_dcgm', 'setup')
         end
 
-        if %w(centos amazon).include?(platform)
+        if %w(centos7 amazon2).include?("#{platform}#{version}")
           it "is not enabled" do
             expect(resource._nvidia_dcgm_enabled).to eq(false)
           end
@@ -164,7 +164,7 @@ def self.setup(chef_run, nvidia_enabled: nil)
             ConvergeNvidiaDcgm.setup(chef_run)
           end
 
-          if %w(centos amazon).include?(platform)
+          if %w(centos7 amazon2).include?("#{platform}#{version}")
             it 'does not install datacenter gpu manager' do
               is_expected.not_to run_bash('Install datacenter-gpu-manager')
             end

@@ -221,21 +221,34 @@ def self.setup(chef_run, nvidia_driver_version: nil)
         end
 
         if platform == 'amazon'
-          compiler_path = version == 2023 ? 'CC=/usr/bin/gcc' : 'CC=/usr/bin/gcc10-gcc'
-          it 'installs gcc10' do
-            is_expected.to install_package('gcc10').with_retries(10).with_retry_delay(5)
+          compiler_version = version == '2023' ? 'gcc' : 'gcc10'
+          compiler_path = version == '2023' ? 'CC=/usr/bin/gcc' : 'CC=/usr/bin/gcc10-gcc'
+          if version == '2'
+            it "installs #{compiler_version}" do
+              is_expected.to install_package(compiler_version).with_retries(10).with_retry_delay(5)
+            end
+            it 'creates dkms/nvidia.conf' do
+              is_expected.to create_template('/etc/dkms/nvidia.conf').with(
+                source: 'nvidia/amazon/dkms/nvidia.conf.erb',
+                cookbook: 'aws-parallelcluster-platform',
+                owner: 'root',
+                group: 'root',
+                mode: '0644',
+                variables: { compiler_path: compiler_path }
+              )
+            end
+          else
+            # Amazon Linux 2023 is expected to install the compiler and create nvidia conf when kernel version is 6.
+            # Here we are testing with kernel version 5
+            it "does not install #{compiler_version}" do
+              is_expected.not_to install_package(compiler_version).with_retries(10).with_retry_delay(5)
+            end
+
+            it 'does not create dkms/nvidia.conf' do
+              is_expected.not_to create_template('/etc/dkms/nvidia.conf')
+            end
           end
 
-          it 'creates dkms/nvidia.conf' do
-            is_expected.to create_template('/etc/dkms/nvidia.conf').with(
-              source: 'nvidia/amazon/dkms/nvidia.conf.erb',
-              cookbook: 'aws-parallelcluster-platform',
-              owner: 'root',
-              group: 'root',
-              mode: '0644',
-              variables: { compiler_path: compiler_path }
-            )
-          end
           it 'installs nvidia driver' do
             is_expected.to run_bash('nvidia.run advanced')
               .with(

@@ -84,7 +84,9 @@ def self.remove(chef_run)
       cached(:arch_suffix) { 'arch_suffix' }
       cached(:nvidia_platform) do
         case platform
-        when 'amazon', 'centos'
+        when 'amazon'
+          version == '2023' ? 'rhel9' : 'rhel7'
+        when 'centos'
           'rhel7'
         when 'redhat', 'rocky'
           "rhel#{version.to_i}"

@@ -45,7 +45,7 @@ def self.setup(chef_run, stunnel_version:, stunnel_checksum:)
         is_expected.to setup_stunnel('setup')
       end
 
-      if platform == 'amazon'
+      if platform == 'amazon' && version == '2'
         it "doesn't install stunnel" do
           is_expected.not_to run_bash('install stunnel')
         end

@@ -40,8 +40,10 @@ def for_oses(os_list)
 def for_all_oses
   [
     %w(amazon 2),
+    %w(amazon 2023),
     %w(ubuntu 20.04),
     %w(ubuntu 22.04),
+    %w(ubuntu 24.04),
     %w(redhat 8),
     %w(rocky 8),
     %w(redhat 9),

@@ -28,10 +28,15 @@ def self.setup(chef_run)
           expect(chef_run).to include_recipe('yum')
         end
 
-        it 'installs epel' do
-          is_expected.to install_alinux_extras_topic('epel')
+        if version == '2'
+          it 'installs epel' do
+            is_expected.to install_alinux_extras_topic('epel')
+          end
+        else
+          it 'does not install epel' do
+            is_expected.not_to install_alinux_extras_topic('epel')
+          end
         end
-
       when 'centos'
         it 'installs yum and epel' do
           expect(chef_run).to include_recipe('yum')

@@ -41,9 +41,9 @@ def self.configure(chef_run)
   for_all_oses do |platform, version|
     context "on #{platform}#{version}" do
       cached(:dns_domain) { 'dns_domain' }
-      cached(:search_domain_config_path) { platform == 'ubuntu' ? '/etc/systemd/resolved.conf' : '/etc/dhcp/dhclient.conf' }
-      cached(:append_pattern) { platform == 'ubuntu' ? 'Domains=*' : 'append domain-name*' }
-      cached(:append_line) { platform == 'ubuntu' ? "Domains=#{dns_domain}" : "append domain-name \" #{dns_domain}\";" }
+      cached(:search_domain_config_path) { platform == 'ubuntu' || platform == 'amazon' && version == '2023' ? '/etc/systemd/resolved.conf' : '/etc/dhcp/dhclient.conf' }
+      cached(:append_pattern) { platform == 'ubuntu' || platform == 'amazon' && version == '2023' ? 'Domains=*' : 'append domain-name*' }
+      cached(:append_line) { platform == 'ubuntu' || platform == 'amazon' && version == '2023' ? "Domains=#{dns_domain}" : "append domain-name \" #{dns_domain}\";" }
 
       cached(:chef_run) do
         runner = runner(platform: platform, version: version, step_into: ['dns_domain']) do |node|
@@ -56,7 +56,7 @@ def self.configure(chef_run)
         is_expected.to configure_dns_domain('configure')
       end
 
-      it 'updates search domaint' do
+      it 'updates search domain' do
         is_expected.to edit_replace_or_add("append Route53 search domain in #{search_domain_config_path}").with(
           path: search_domain_config_path,
           pattern: append_pattern,

@@ -30,6 +30,8 @@ def self.validate(chef_run)
         cached(:package_platform) do
           platform_version = if version.to_i == 2
                                7
+                             elsif platform == 'amazon' && version == '2023'
+                               9
                              else
                                version.to_i
                              end