[DONOTMERGE][B200] Add debugging commands.

Author: gmarciani

cookbooks/aws-parallelcluster-platform/recipes/install.rb

@@ -14,9 +14,36 @@
 include_recipe "aws-parallelcluster-platform::sudo_install"
 include_recipe "aws-parallelcluster-platform::users"
 include_recipe "aws-parallelcluster-platform::disable_services"
+bash "Check kernel before repo setup" do
+  code <<-CODE
+    set -x
+    uname -a
+    yum info infiniband-diags libibumad 
+    yum provides infiniband-diags libibumad 
+    yum install --assumeno infiniband-diags libibumad
+    CODE
+end
 package_repos 'setup the repositories'
+bash "Check kernel after repo setup" do
+  code <<-CODE
+    set -x
+    uname -a
+    yum info infiniband-diags libibumad 
+    yum provides infiniband-diags libibumad
+    yum install --assumeno infiniband-diags libibumad
+    CODE
+end
 include_recipe "aws-parallelcluster-platform::directories"
 install_packages 'Install OS and extra packages'
+bash "Check kernel after extra packages" do
+  code <<-CODE
+    set -x
+    uname -a
+    yum info infiniband-diags libibumad 
+    yum provides infiniband-diags libibumad
+    yum install --assumeno infiniband-diags libibumad 
+    CODE
+end
 include_recipe "aws-parallelcluster-platform::cookbook_virtualenv"
 include_recipe "aws-parallelcluster-platform::awscli"
 unless alinux2023_on_docker? # Running this recipe on Alinux 2023 docker generates false failure.

cookbooks/aws-parallelcluster-platform/resources/nvidia_nvlsm/nvidia_nvlsm_alinux2023.rb

@@ -22,3 +22,7 @@
 def platform
   "amzn#{node['platform_version'].to_i}"
 end
+
+def nvidia_nvlsm_dependencies
+  %(infiniband-diags libibumad)
+end

cookbooks/aws-parallelcluster-platform/resources/nvidia_nvlsm/partial/_nvidia_nvlsm_debian.rb

@@ -34,9 +34,10 @@ def nvidia_nvlsm_dependencies
   #  because otherwise the kernel pinning would block the installation. Once we figure out how to make this installation
   #  work with kernel pinning, we will move the installation here.
   # %(linux-modules-extra-aws infiniband-diags ibutils)
+  %(infiniband-diags ibutils)
   []
 end
 
 def nvidia_nvlsm_install_preconditions_commands
-  "apt-cache policy ; apt-mark showhold"
+  "uname -a ; apt-cache policy ; apt-mark showhold"
 end

cookbooks/aws-parallelcluster-platform/resources/nvidia_nvlsm/partial/_nvidia_nvlsm_rhel.rb

@@ -33,10 +33,9 @@ def nvidia_nvlsm_dependencies
   #  The installation has been temporarily moved to the ParallelCluster component, before cookbook execution,
   #  because otherwise the kernel pinning would block the installation. Once we figure out how to make this installation
   #  work with kernel pinning, we will move the installation here.
-  # %(kernel-modules-extra-aws infiniband-diags libibumad)
-  []
+  %(infiniband-diags libibumad)
 end
 
 def nvidia_nvlsm_install_preconditions_commands
-  "yum repolist all ; yum versionlock list"
+  "uname -a ; yum repolist all ; yum versionlock list ; yum info kernel-modules-extra-aws infiniband-diags libibumad ; yum provides kernel-modules-extra-aws infiniband-diags libibumad"
 end