added ansible for deployment in private infrastrucutre

deploy/roles/cuda/README.md

@@ -0,0 +1,23 @@
+# The `cuda` Role
+
+The `cuda` role installs the Nvidia CUDA GPU repository and drivers.
+
+This role may be executed independently by running:
+
+```bash
+$ ansible-playbook packages.yml --tags "cuda"
+```
+
+On execution, this role installs the following directly to the remote machine's operating system:
+
+### Apt Repositories (Debian/Ubuntu)
+
+- Nvidia CUDA
+
+### Apt GPG Signing Keys (Debian/Ubuntu)
+
+- cudatools <cudatools@nvidia.com>
+
+### Operating System Packages (Debian/Ubuntu)
+
+- cuda v10.x

deploy/roles/cuda/handlers/main.yml

@@ -0,0 +1,5 @@
+---
+# CUDA Driver Setup
+- name: reboot the machine
+  reboot:
+  listen: "reboot the machine"

deploy/roles/cuda/tasks/debian.yml

@@ -0,0 +1,22 @@
+---
+# Cuda Driver Package Installation
+- name: copy the CUDA repo onto the machine
+  copy:
+    src: "{{ cuda.apt.repo }}"
+    dest: "{{ general.working_dir }}/{{ cuda.apt.repo }}"
+
+- name: get the CUDA signing key
+  apt_key:
+    state: present
+    url: "{{ cuda.apt.signing_key_url }}"
+
+- name: add the CUDA repo
+  apt:
+    deb: "{{ general.working_dir }}/{{ cuda.apt.repo }}"
+    state: present
+
+- name: install the CUDA drivers
+  apt:
+    name: "{{ cuda.apt.package }}"
+    update_cache: yes
+  notify: "reboot the machine"

deploy/roles/cuda/tasks/main.yml

@@ -0,0 +1,3 @@
+---
+- import_tasks: debian.yml
+  when: ansible_facts['os_family']|lower == 'debian' # debian, ubuntu

deploy/roles/cuda/vars/main.yml

@@ -0,0 +1,9 @@
+---
+cuda:
+  apt:
+    package: cuda=10.* # known good with 10.0.130-1
+    repo: cuda-repo-ubuntu1804_10.0.130-1_amd64.deb # Network install
+    signing_key_url: https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/7fa2af80.pub
+
+general:
+  working_dir: /tmp

deploy/roles/docker/README.md

@@ -0,0 +1,33 @@
+# The `docker` Role
+
+The `docker` role installs the Docker CE repository and runtime packages.
+
+This role may be executed independently by running:
+
+```bash
+$ ansible-playbook packages.yml --tags "docker"
+```
+
+On execution, this role installs the following directly to the remote machine's operating system:
+
+### Apt Repositories (Debian/Ubuntu)
+
+- Docker CE
+
+### Apt GPG Signing Keys (Debian/Ubuntu)
+
+- Docker Release (CE deb) <docker@docker.com>
+
+### Operating System Packages (Debian/Ubuntu)
+
+- docker-ce v5:18.x
+
+### System Services
+
+- `docker` (enabled and started)
+
+### User Groups
+
+- docker
+
+All users listed under `secrets.users` in `secrets.yml` will be added to the `docker` group.

deploy/roles/docker/tasks/debian.yml

@@ -0,0 +1,17 @@
+---
+# Docker Package Installation
+- name: add Docker's official GPG key
+  apt_key:
+    state: present
+    url: "{{ docker.apt.signing_key_url }}"
+
+- name: add Docker apt repository
+  apt_repository:
+    repo: "{{ docker.apt.repo }}"
+    state: present
+
+- name: install Docker CE
+  apt:
+    name: "{{ docker.apt.package }}"
+    state: present
+    update_cache: yes

deploy/roles/docker/tasks/main.yml

@@ -0,0 +1,23 @@
+---
+# General Docker Setup
+- import_tasks: debian.yml
+  when: ansible_facts['os_family']|lower == 'debian' # debian, ubuntu
+
+- name: ensure the docker service starts at boot
+  systemd:
+    name: "{{ docker.service_name }}"
+    enabled: yes
+    state: started
+
+- name: create Docker group to avoid running as root
+  group:
+    name: "{{ docker.group_name }}"
+    state: present
+
+- name: add users to Docker group
+  user:
+    append: yes
+    groups:
+      - "{{ docker.group_name }}"
+    name: "{{ item }}"
+  loop: "{{ users }}"

deploy/roles/docker/vars/main.yml

@@ -0,0 +1,8 @@
+---
+docker:
+  apt:
+    package: docker-ce=5:18.* # known good with 5:18.09.1~3-0~ubuntu-bionic
+    repo: deb [arch=amd64] https://download.docker.com/linux/ubuntu bionic stable
+    signing_key_url: https://download.docker.com/linux/ubuntu/gpg
+  group_name: docker
+  service_name: docker

deploy/roles/nvidia/README.md

@@ -0,0 +1,31 @@
+# The `nvidia` Role
+
+The `nvidia` role installs the Nvidia Docker runtime, enabling access to Nvidia CUDA GPU hardware from within Docker containers.
+
+This role may be executed independently by running:
+
+```bash
+$ ansible-playbook packages.yml --tags "nvidia"
+```
+
+On execution, this role installs the following directly to the remote machine's operating system:
+
+### Apt Repositories (Debian/Ubuntu)
+
+- libnvidia_container
+- nvidia_container_runtime
+- nvidia_docker
+
+### Apt GPG Signing Keys (Debian/Ubuntu)
+
+- NVIDIA CORPORATION (Open Source Projects) <cudatools@nvidia.com>
+
+### Operating System Packages (Debian/Ubuntu)
+
+- nvidia-docker2 v2.x
+
+Containers are launched with the Nvidia Docker runtime by passing the `--runtime=nvidia` flag, e.g.,
+
+```bash
+$ docker run --rm --runtime=nvidia nvidia/cuda nvidia-smi
+```

deploy/roles/nvidia/handlers/main.yml

@@ -0,0 +1,6 @@
+---
+- name: reload the docker service
+  systemd:
+    name: docker
+    state: reloaded
+  listen: "reload the docker service"

deploy/roles/nvidia/tasks/debian.yml

@@ -0,0 +1,31 @@
+---
+# Nvidia Docker Runtime Installation
+- name: get the NVIDIA signing key
+  apt_key:
+    state: present
+    url: "{{ nvidia.apt.signing_key_url }}"
+
+- name: add the libnvidia-container repo
+  apt_repository:
+    filename: "{{ nvidia.apt.repo_filename }}"
+    repo: "{{ nvidia.apt.repos.libnvidia_container }}"
+    state: present
+
+- name: add the nvidia-container-runtime repo
+  apt_repository:
+    filename: "{{ nvidia.apt.repo_filename }}"
+    repo: "{{ nvidia.apt.repos.nvidia_container_runtime }}"
+    state: present
+
+- name: add the nvidia-docker repo
+  apt_repository:
+    filename: "{{ nvidia.apt.repo_filename }}"
+    repo: "{{ nvidia.apt.repos.nvidia_docker }}"
+    state: present
+
+- name: install the nvidia-docker2 runtime
+  apt:
+    name: "{{ nvidia.apt.package }}"
+    state: present
+    update_cache: yes
+  notify: "reload the docker service"

deploy/roles/nvidia/tasks/main.yml

@@ -0,0 +1,4 @@
+---
+# Nvidia Docker Runtime Setup
+- import_tasks: debian.yml
+  when: ansible_facts['os_family']|lower == 'debian' # debian, ubuntu

deploy/roles/nvidia/vars/main.yml

@@ -0,0 +1,10 @@
+---
+nvidia:
+  apt:
+    package: nvidia-docker2=2.* # known good with 2.0.3+docker18.09.1-1
+    repo_filename: nvidia_github_io_nvidia_docker_ubuntu18_04_nvidia_docker.list
+    repos:
+      libnvidia_container: deb https://nvidia.github.io/libnvidia-container/ubuntu18.04/$(ARCH) /
+      nvidia_container_runtime: deb https://nvidia.github.io/nvidia-container-runtime/ubuntu18.04/$(ARCH) /
+      nvidia_docker: deb https://nvidia.github.io/nvidia-docker/ubuntu18.04/$(ARCH) /
+    signing_key_url: https://nvidia.github.io/nvidia-docker/gpgkey

deploy/roles/packages/README.md

@@ -0,0 +1,26 @@
+# The `packages` Role
+
+The `packages` role installs the operating system and Python pre-requisites that are required to execute the other roles.
+
+This role may be executed independently by running:
+
+```bash
+$ ansible-playbook packages.yml --tags "packages"
+```
+
+On execution, this role installs the following directly to the remote machine's operating system:
+
+### Operating System Packages (Debian/Ubuntu)
+
+- apt-transport-https v1.x
+- ca-certificates v20180409
+- curl v7.x
+- gnupg-agent v2.x
+- python-apt v1.x
+- python-pip v9.x
+- python-setuptools v39.x
+- software-properties-common v0.96.24.32.7
+
+### Python (pip) Packages
+
+- docker v3.7.0 or greater

deploy/roles/packages/tasks/debian.yml

@@ -0,0 +1,12 @@
+---
+# General Dependencies
+- name: install general apt dependencies
+  apt:
+    name: "{{ general.dependencies.apt }}"
+    state: present
+    update_cache: yes
+
+- name: install general pip dependencies for Ansible
+  pip:
+    name: "{{ general.dependencies.pip }}"
+    state: present

deploy/roles/packages/tasks/main.yml

@@ -0,0 +1,3 @@
+---
+- import_tasks: debian.yml
+  when: ansible_facts['os_family']|lower == 'debian' # debian, ubuntu

deploy/roles/packages/vars/main.yml

@@ -0,0 +1,17 @@
+---
+# General Dependencies
+general:
+  dependencies:
+    apt:
+      - apt-transport-https=1.* # known good with 1.6.6ubuntu0.1 all
+      - ca-certificates=20180409 # known good with 20180409
+      - curl=7.* # known good with 7.58.0-2ubuntu3.5
+      - gnupg-agent=2.* # known good with 2.2.4-1ubuntu1.2
+      - python3-apt=1.* # known good with 1.6.3
+      - python3-pip=9.* # known good with 9.0.1-2.3~ubuntu1
+      - python3-setuptools=39.* # known good with 39.0.1-2
+      - software-properties-common=0.96.24.32.7 # known good with 0.96.24.32.7
+      - git
+      - golang
+    pip:
+      - docker>=3.7.0 # known good with 3.7.0

deploy/roles/rsync/files/rsyncd.example.conf

@@ -0,0 +1,16 @@
+motd file = /etc/rsyncd.motd
+log file = /var/log/rsyncd.log
+pid file = /var/run/rsyncd.pid
+lock file = /var/run/rsync.lock
+
+[jupyter_data]
+   path = /nfs/jupyter/data
+   comment = Jupyter data files
+   uid = nobody
+   gid = nobody
+   read only = yes
+   list = yes
+   auth users = @rsync:ro
+   secrets file = /etc/rsyncd.secrets
+   hosts allow = 192.168.1.0/24
+   hosts deny = 0.0.0.0/0

deploy/roles/rsync/handlers/main.yml

@@ -0,0 +1,5 @@
+---
+- name: restart the rsync service
+  systemd:
+    name: "{{ rsync.service_name }}"
+    state: restarted

deploy/roles/rsync/tasks/main.yml

@@ -0,0 +1,26 @@
+---
+# Configure the rsync daemon on the master
+- name: copy rsync.conf up to the server
+  copy:
+    src: "{{ rsync.config_file.src }}"
+    dest: "{{ rsync.config_file.dest }}"
+  notify:
+    - restart the rsync service
+  when: inventory_hostname == rsync.master
+
+- name: configure rsync on the server
+  service:
+    enabled: yes
+    name: "{{ rsync.service_name }}"
+    state: started
+  when: inventory_hostname == rsync.master
+
+# Synchronize the data folder across hosts
+- name: synchronize the data folder
+  delegate_to: "{{ rsync.master }}"
+  synchronize:
+    src: "{{ sync.source }}"
+    dest: "{{ sync.destination }}"
+    rsync_opts: "{{ rsync.options }}"
+    use_ssh_args: yes
+  when: inventory_hostname != rsync.master

deploy/roles/rsync/vars/main.yml

@@ -0,0 +1,10 @@
+---
+rsync:
+  config_file:
+    src: rsyncd.conf
+    dest: /etc/rsyncd.conf
+  master: ml1
+  options:
+    - "-e ssh"
+    - "--no-motd"
+  service_name: rsync