Release updates

config.example/group_vars/all.yml

@@ -122,7 +122,9 @@ sftp_chroot: false
 ################################################################################
 # NVIDIA GPU configuration
 # Playbook: nvidia-cuda
-cuda_version: cuda-toolkit-11-5
+# Install latest version by default,
+# if you want a specific version, use i.e. cuda-toolkit=12.2.0-1
+# cuda_version: cuda-toolkit
 
 # DGX-specific vars may be used to target specific models,
 # because available versions for DGX may differ from the generic repo
@@ -146,9 +148,9 @@ nvidia_driver_force_install: false
 # Docker configuration
 # Playbook: docker, nvidia-docker, k8s-cluster
 #
-# For supported Docker versions, see: kubespray/roles/container-engine/docker/vars/*
+# For supported Docker versions, see: submodules/kubespray/roles/container-engine/docker/vars/*
 docker_install: yes
-docker_version: '20.10'
+# docker_version: 'latest'
 docker_dns_servers_strict: no
 docker_storage_options: -s overlay2
 #docker_options: "--bip=192.168.99.1/24"
@@ -196,7 +198,9 @@ enroot_environ_config_files_dgx:
 # Singularity configuration
 # Playbook: singularity, slurm-cluster
 # Singularity target version
-singularity_version: "3.7.3"
+# set an alternate singularity version here;
+# see roles/singularity_wrapper/defaults/main.yml for default
+# singularity_version:
 singularity_conf_path: "/etc/singularity/singularity.conf"
 bind_paths: []
   # example:
@@ -275,10 +279,6 @@ deepops_dir: /opt/deepops
 # Roles: K8s GPU operator, GPU plugin, OpenShift/K8s
 deepops_venv: '{{ deepops_dir }}/venv'
 
-# OpenMPI
-# Playbook: openmpi
-openmpi_version: 4.0.3
-
 # Disable cloud-init
 deepops_disable_cloud_init: true
 

config.example/group_vars/slurm-cluster.yml

@@ -3,7 +3,7 @@
 ################################################################################
 # Slurm job scheduler configuration
 # Playbook: slurm, slurm-cluster, slurm-perf, slurm-perf-cluster, slurm-validation
-slurm_version: "22.05.2"
+# slurm_version: ""
 slurm_install_prefix: /usr/local
 pmix_install_prefix: /opt/deepops/pmix
 hwloc_install_prefix: /opt/deepops/hwloc
@@ -137,10 +137,10 @@ sm_install_host: "slurm-master[0]"
 slurm_install_hpcsdk: true
 
 # Select the version of HPC SDK to download
-hpcsdk_major_version: "22"
-hpcsdk_minor_version: "1"
-hpcsdk_file_cuda: "11.5"
-hpcsdk_arch: "x86_64"
+#hpcsdk_major_version: ""
+#hpcsdk_minor_version: ""
+#hpcsdk_file_cuda: ""
+#hpcsdk_arch: "x86_64"
 
 # In a Slurm cluster, default to setting up HPC SDK as modules rather than in
 # the default user environment
@@ -156,7 +156,7 @@ hpcsdk_install_in_path: false
 # this can help you get started.
 ################################################################################
 slurm_cluster_install_openmpi: false
-openmpi_version: 4.0.4
+#openmpi_version:
 openmpi_install_prefix: "/usr/local"
 openmpi_configure: "./configure --prefix={{ openmpi_install_prefix }} --disable-dependency-tracking --disable-getpwuid --with-pmix={{ pmix_install_prefix }} --with-hwloc={{ hwloc_install_prefix }} --with-pmi={{ slurm_install_prefix }} --with-slurm={{ slurm_install_prefix }} --with-libevent=/usr"
 
@@ -185,7 +185,7 @@ allow_user_set_gpu_clocks: no
 ################################################################################
 slurm_install_enroot: true
 slurm_install_pyxis: true
-slurm_pyxis_version: 0.11.1
+#slurm_pyxis_version:
 
 # /run is default partition of pyxis runtime_path
 resize_run_partition: false

roles/nhc/defaults/main.yml

@@ -1,5 +1,5 @@
 ---
-nhc_version: "1.4.2"
+nhc_version: "1.4.3"
 nhc_src_url: "https://github.com/mej/nhc/releases/download/{{ nhc_version }}/lbnl-nhc-{{ nhc_version }}.tar.xz"
 nhc_install_dir: "/usr"
 nhc_config_dir: "/etc"

roles/nhc/molecule/default/molecule.yml

@@ -16,6 +16,12 @@ platforms:
       - /sys/fs/cgroup:/sys/fs/cgroup:ro
     privileged: true
     pre_build_image: true
+  - name: nhc-ubuntu-2204
+    image: geerlingguy/docker-ubuntu2204-ansible
+    volumes:
+      - /sys/fs/cgroup:/sys/fs/cgroup:ro
+    privileged: true
+    pre_build_image: true
   - name: nhc-centos-7
     image: geerlingguy/docker-centos7-ansible
     volumes:

roles/nhc/vars/ubuntu-22.04.yml

@@ -0,0 +1,5 @@
+---
+nhc_build_deps:
+  - build-essential
+
+nhc_ssh_daemon: "sshd:"

roles/nvidia_cuda/defaults/main.yml

@@ -1,6 +1,7 @@
 ---
-# 'cuda' is the generic package and will pull the latest version
-cuda_version: "cuda-toolkit-11-7"
+# 'cuda-toolkit' is the generic package and will pull the latest version
+# if you want a specific version, use i.e. cuda-toolkit=12.2.0-1
+cuda_version: cuda-toolkit
 
 # DGX-specific vars may be used to target specific models,
 # because available versions for DGX may differ from the generic repo

roles/nvidia_hpc_sdk/defaults/main.yml

@@ -15,15 +15,15 @@
 # See https://developer.nvidia.com/nvidia-hpc-sdk-downloads for more detail on available downloads.
 
 # Version strings used to construct download URL
-hpcsdk_major_version: "22"
-hpcsdk_minor_version: "1"
-hpcsdk_file_cuda: "11.5"
+hpcsdk_major_version: "23"
+hpcsdk_minor_version: "7"
+hpcsdk_file_cuda: "12.2"
 hpcsdk_arch: "x86_64"
 
 # We need to specify the default CUDA toolkit to use during installation.
 # This should usually be the latest CUDA included in the HPC SDK you are
 # installing.
-hpcsdk_default_cuda: "11.5"
+hpcsdk_default_cuda: "12.2"
 
 # Add HPC SDK modules to the MODULEPATH?
 hpcsdk_install_as_modules: false

roles/singularity_wrapper/defaults/main.yml

@@ -1,6 +1,6 @@
 ---
 # vars for lecorguille.singularity
-singularity_version: "3.7.3"
+singularity_version: "3.11.4"
 singularity_conf_path: "/etc/singularity/singularity.conf"
 bind_paths: []
 

roles/slurm/defaults/main.yml

@@ -7,7 +7,7 @@ hwloc_build_dir: /opt/deepops/build/hwloc
 pmix_build_dir: /opt/deepops/build/pmix
 
 slurm_workflow_build: yes
-slurm_version: "22.05.2"
+slurm_version: "23.02.4"
 slurm_src_url: "https://download.schedmd.com/slurm/slurm-{{ slurm_version }}.tar.bz2"
 slurm_build_make_clean: no
 slurm_build_dir_cleanup: no

roles/slurm/vars/ubuntu.yml

@@ -5,8 +5,9 @@ slurm_build_deps:
   - build-essential
   - libmunge-dev
   - libmariadb-dev
-  - libmariadbclient-dev
+  - libmariadbclient-dev-compat
   - libpam0g-dev
+  - libdbus-1-dev
   - python3-minimal
   - ruby-dev
   - wget