Feature azure

compute.yml

@@ -14,7 +14,7 @@
     - citc_user
     - filesystem
     - ssh
-    #- security_updates
+    ##- security_updates
     - ntp
     - sssd
     - lmod

group_vars/compute.yml

@@ -14,6 +14,9 @@ mpi_packages:
   google:
     - mpich
     - openmpi
+  azure:
+    - mpich
+    - openmpi
   aws: []
 
 monitoring_role: client

group_vars/management.yml

@@ -12,6 +12,8 @@ slurm_role: mgmt
 slurm_elastic:
   oracle:
     config_directory: /home/slurm/.oci/
+  azure:
+    config_directory: /home/slurm/.oci/
 
 install_packages:
   - xorg-x11-xauth
@@ -27,6 +29,9 @@ mpi_packages:
   google:
     - openmpi-devel
     - mpich-devel
+  azure:
+    - openmpi-devel
+    - mpich-devel
   aws: []
 
 monitoring_role: master

roles/filesystem/tasks/main.yml

@@ -26,7 +26,18 @@
     opts: defaults,nofail,nosuid
     state: mounted
   when:
-    ansible_local.citc.csp != "aws"
+    - ansible_local.citc.csp != "aws"
+    - ansible_local.citc.csp != "azure"
+
+- name: Mount shared file system now that fileserver is ready
+  mount:
+    path: /mnt/{{ filesystem_mount_point }}
+    src: "{{ filesystem_target_address }}:{{ filesystem_mount_point }}"
+    fstype: nfs
+    opts: rw,hard,rsize=1048576,wsize=1048576,vers=3,tcp,_netdev,noauto
+    state: mounted
+  when:
+    - ansible_local.citc.csp == "azure"
 
 - name: Mount shared file system
   mount:

roles/finalise/tasks/main.yml

@@ -9,6 +9,12 @@
   delay: 10
   tags: packer
 
+- name: update directory mode for the finalised files
+  file:
+    path: /mnt/shared
+    state: directory
+    mode: 0755
+
 - name: create directory for the finalised files
   file:
     path: /mnt/shared/finalised

roles/monitoring/tasks/main.yml

@@ -20,11 +20,16 @@
     baseurl: https://repos.influxdata.com/centos/$releasever/{{ "arm64" if ansible_architecture == "aarch64" else ansible_architecture }}/stable/
     gpgkey: https://repos.influxdata.com/influxdb.key
 
+#- name: install telegraf package
+#  package:
+#    name: telegraf
+#    state: present
+#  notify: restart telegraf
+
 - name: install telegraf package
-  package:
+  yum:
     name: telegraf
-    state: present
-  notify: restart telegraf
+    disable_gpg_check: yes
 
 - name: enable the telegraf service
   service:

roles/packer/files/all.pkr.hcl

@@ -10,6 +10,13 @@ variable "aws_region" {}
 variable "aws_instance_type" {}
 variable "aws_arch" {}
 
+variable "azure_region" {}
+variable "azure_instance_type" {}
+variable "azure_resource_group" {}
+variable "azure_virtual_network" {}
+variable "azure_virtual_network_subnet" {}
+variable "azure_dns_zone" {}
+
 variable "oracle_availability_domain" {}
 variable "oracle_base_image_ocid" {}
 variable "oracle_compartment_ocid" {}
@@ -91,6 +98,22 @@ source "amazon-ebs" "aws" {
     }
 }
 
+source "azure-arm" "azure" {
+    managed_image_name = "${var.destination_image_name}-${var.cluster}-v{{timestamp}}"
+    managed_image_resource_group_name = var.azure_resource_group
+    build_resource_group_name = var.azure_resource_group
+    virtual_network_name = var.azure_virtual_network
+    virtual_network_subnet_name = var.azure_virtual_network_subnet
+    virtual_network_resource_group_name = var.azure_resource_group
+    vm_size = var.azure_instance_type
+    ssh_username = var.ssh_username
+    os_type = "Linux"
+    image_publisher = "OpenLogic"
+    image_offer = "CentOS"
+    image_sku = "8_4-gen2"
+}
+
+
 source "oracle-oci" "oracle" {
     image_name = "${var.destination_image_name}-${var.cluster}-v{{timestamp}}"
     availability_domain = var.oracle_availability_domain
@@ -111,6 +134,7 @@ build {
         "source.googlecompute.google",
         "source.amazon-ebs.aws",
         "source.oracle-oci.oracle",
+        "source.azure-arm.azure",
     ]
 
     provisioner "file" {
@@ -161,4 +185,12 @@ build {
     provisioner "shell" {
         script = "/home/citc/compute_image_extra.sh"
     }
+
+    provisioner "shell" {
+        script = "/home/citc/install_cvmfs_eessi.sh"
+    }
+
+    provisioner "shell" {
+        script = "/home/citc/compute_image_finalize.sh"
+    }
 }

roles/packer/files/compute_image_extra.sh

@@ -8,4 +8,4 @@
 # sudo yum -y install cmake gcc-gfortran
 
 # to install CernVM-FS and configure access to EESSI, uncomment the line below:
-# /home/citc/install_cvmfs_eessi.sh
+#/home/citc/install_cvmfs_eessi.sh

roles/packer/files/compute_image_finalize.sh

@@ -0,0 +1,3 @@
+#! /bin/bash
+
+/usr/sbin/waagent -force -deprovision && export HISTSIZE=0 && sync

roles/packer/tasks/main.yml

@@ -65,6 +65,14 @@
     group: citc
     mode: u=rw,g=rw,o=
 
+- name: copy in packer finalize run script template
+  copy:
+    src: compute_image_finalize.sh
+    dest: /home/citc/compute_image_finalize.sh
+    owner: citc
+    group: citc
+    mode: u=rw,g=rw,o=
+
 - name: copy in EESSI install script
   copy:
     src: install_cvmfs_eessi.sh

roles/packer/templates/prepare_ansible.sh.j2

@@ -9,14 +9,24 @@ $(hostname)
 cluster_id={{ startnode_config.cluster_id }}
 packer_run=yes
 EOF'
-{% if ansible_local.citc.csp in ["aws", "google"] %}
+{% if ansible_local.citc.csp in ["aws", "google", "azure"] %}
 sudo yum install -y epel-release
 sudo dnf config-manager --set-enabled powertools
 {% elif ansible_local.citc.csp == "oracle" %}
 sudo dnf install -y oracle-epel-release-el8
 sudo dnf install -y https://dl.fedoraproject.org/pub/epel/epel-release-latest-8.noarch.rpm
 sudo dnf config-manager --set-enabled ol8_codeready_builder
 {% endif %}
+{% if ansible_local.citc.csp in ["azure"] %}
+echo "[main]" | sudo tee -a /etc/NetworkManager/conf.d/dhclient.conf
+echo "dhcp=dhclient" | sudo tee -a /etc/NetworkManager/conf.d/dhclient.conf
+cat /etc/NetworkManager/conf.d/dhclient.conf
+echo "append domain-name \" {{ startnode_config.dns_zone }}\";" | sudo tee -a /etc/dhcp/dhclient.conf
+echo "append domain-search \" {{ startnode_config.dns_zone }}\";" | sudo tee -a /etc/dhcp/dhclient.conf
+sudo cat /etc/dhcp/dhclient.conf
+sudo systemctl restart NetworkManager
+cat /etc/resolv.conf
+{% endif %}
 sudo dnf install -y ansible git
 sudo cat /tmp/hosts
 sudo mkdir -p /etc/ansible/facts.d/

roles/packer/templates/variables.pkrvars.hcl.j2

@@ -1,12 +1,19 @@
 ca_cert = "{{ ca_cert }}"
 cluster = "{{ startnode_config.cluster_id }}"
 destination_image_name = "citc-slurm-compute"
-ssh_username = "{%- if ansible_local.citc.csp in ["aws", "google"] -%}centos{%- else -%}opc{%- endif -%}"
+ssh_username = "{%- if ansible_local.citc.csp in ["aws", "google", "azure"] -%}centos{%- else -%}opc{%- endif -%}"
 
 aws_arch = "x86_64"
 aws_region = "{%- if startnode_config.region is defined -%}{{ startnode_config.region }}{%- endif -%}"
 aws_instance_type = "t2.nano"
 
+azure_region = "{%- if startnode_config.region is defined -%}{{ startnode_config.region }}{%- endif -%}"
+azure_resource_group = "{%- if startnode_config.resource_group is defined -%}{{ startnode_config.resource_group }}{%- endif -%}"
+azure_virtual_network = "{%- if startnode_config.virtual_network is defined -%}{{ startnode_config.virtual_network }}{%- endif -%}"
+azure_virtual_network_subnet = "{%- if startnode_config.virtual_network_subnet is defined -%}{{ startnode_config.virtual_network_subnet }}{%- endif -%}"
+azure_dns_zone = "{%- if startnode_config.dns_zone is defined -%}{{ startnode_config.dns_zone }}{%- endif -%}"
+azure_instance_type = "Standard_D4s_v3"
+
 google_destination_image_family = "citc-slurm-compute"
 google_network = "{%- if startnode_config.network_name is defined -%}{{ startnode_config.network_name }}{%- endif -%}"
 google_source_image_family = "centos-8"