1. Introduction
2. Requirements
3. Download
4. Configuration
5. Packer Machine Image Builds
6. Troubleshoot
7. Known Issues
8. Unsupported Features
9. Contributing
10. Credits

This repository provides opinionated infrastructure-as-code examples to automate the creation of virtual machine images and their guest operating systems on Proxmox using HashiCorp Packer and the Packer Plugin for Proxmox (proxmox-iso builder). All examples are authored in the HashiCorp Configuration Language ("HCL2").

By default, the machine image artifacts are converted to templates within Proxmox after a virtual machine is built and configured according to the individual templates.

The following builds are available:

Operating Systems:

Operating systems and versions tested with the project:

• Proxmox PVE Version >= 8
• Ubuntu Server 22.04 LTS (x86_64)
• CentOS Stream 9 (x86_64)

Packer:

• HashiCorp Packer 1.12.0 or higher.

  Note

  Click on the operating system name to display the installation steps.

  • Ubuntu

    The Terraform packages are signed using a private key controlled by HashiCorp, so you must configure your system to trust that HashiCorp key for package authentication.

    To configure your repository:

    sudo bash -c 'wget -O- https://apt.releases.hashicorp.com/gpg | gpg --dearmor > /usr/share/keyrings/hashicorp-archive-keyring.gpg'

    Verify the key's fingerprint:

    gpg --no-default-keyring --keyring /usr/share/keyrings/hashicorp-archive-keyring.gpg --fingerprint

    The fingerprint must match E8A0 32E0 94D8 EB4E A189 D270 DA41 8C88 A321 9F7B. You can also verify the key on [Security at HashiCorp][hcp-security] under Linux Package Checksum Verification.

    Add the official HashiCorp repository to your system:

    sudo bash -c 'echo "deb [signed-by=/usr/share/keyrings/hashicorp-archive-keyring.gpg] \
    https://apt.releases.hashicorp.com $(lsb_release -cs) main" > /etc/apt/sources.list.d/hashicorp.list'

    Install Packer from HashiCorp repository:

    sudo apt update && sudo apt install packer

  • CentOS Stream 9

    Install yum-config-manager to manage your repositories.

    sudo yum install -y yum-utils

    Use yum-config-manager to add the official HashiCorp Linux repository:

    sudo yum-config-manager --add-repo https://rpm.releases.hashicorp.com/RHEL/hashicorp.repo

    Install.

    sudo yum -y install packer

• Packer plugins:

  Note

  Required plugins are automatically downloaded and initialized when using ./build.sh. For dark sites, you may download the plugins and place these same directory as your Packer executable /usr/local/bin or $HOME/.packer.d/plugins.

  • Packer Plugin for Git 0.6.2 or later - a community plugin for HashiCorp Packer.
  • HashiCorp Packer Plugin for Proxmox-ISO version 1.2.1 - the plugin for HashiCorp Packer to communicate with Proxmox VE. This needs to be pinned to version 1.2.1 at this time due to a CPU bug.

Ansible:

• Ansible Core version 2.14 or higher.

  Note

  Click on the operating system name to display the installation steps.

  • Ubuntu

    It is recommended that you install ansible-core using your system's package manager instead of via pip.

    Refresh the repositories:

    sudo apt update

    Install software-properties-common:

    sudo apt install -y software-properties-common

    Add the Ansible repository to your system:

    sudo add-apt-repository --yes --update ppa:ansible/ansible

    Install ansible-core from the Ansible repository:

    sudo apt install -y ansible-core

  • CentOS Stream 9

    It is recommended that you install ansible using your system's package manager instead of via pip.

    Install.

    dnf -y install ansible

Platform:

• Proxmox PVE 8.0 or later.

After installing the required software, the quickest way to get building is to clone this repository.

git clone https://github.com/ajschroeder/packer-examples-for-proxmox.git
cd packer-examples-for-proxmox

The following table describes the directory structure.

The project includes example variables files that you can use as a starting point for your own configuration.

The variables are defined in .pkrvars.hcl files.

Run the config script ./config.sh to copy the .pkrvars.hcl.example files to a config directory.

./config.sh
./build.sh

The config folder is the default folder. You can override the default by passing an alternate value as the first argument.

For example:

San Francisco: us-west-1

./config.sh us-west-1
./build.sh us-west-1

Los Angeles: us-west-2

./config.sh us-west-2
./build.sh us-west-2

This is useful for the purposes of running machine image builds for different environment.

Edit the config/ansible.pkrvars.hcl file to configure the credentials for the Ansible account on Linux machine images.

ansible_username = "ansible"
ansible_key      = "<public_key>"

Ansible User Password

A random password is auto-generated for the Ansible user.

Edit the config/build.pkrvars.hcl file to configure the credentials for the default account on machine images.

build_username           = "example"
build_password           = "<plaintext_password>"
build_password_encrypted = "<sha512_encrypted_password>"
build_key                = "<public_key>"

You will need to generate a SHA-512 encrypted password for the build_password_encrypted using tools like mkpasswd.

Run the following command to generate a SHA-512 encrypted password:

mkpasswd -m sha512crypt

The following output is displayed:

Password: ***************
[password hash]

Generate a public key for the build_key for public key authentication.

Run the following command to generate a public key for the build_key for public key authentication.

ssh-keygen -t ecdsa -b 512 -C "<name@example.com>"

The following output is displayed:

Generating public/private ecdsa key pair.
Enter file in which to save the key (/Users/example/.ssh/id_ecdsa):
Enter passphrase (empty for no passphrase): **************
Enter same passphrase again: **************
Your identification has been saved in /Users/example/.ssh/id_ecdsa.
Your public key has been saved in /Users/example/.ssh/id_ecdsa.pub.

The content of the public key, build_key, is added the key to the ~/.ssh/authorized_keys file of the build_username on the Linux guest operating systems.

Edit the config/common.pkrvars.hcl file to configure the following common variables:

• Removable Media Settings
• Boot and Provisioning Settings
• HCP Packer Registry

// Removable Media Settings
common_iso_storage = "<Proxmox Storage Location>"

// Boot and Provisioning Settings
common_data_source       = "http"
common_http_interface    = null
common_http_bind_address = null
common_http_port_min     = 8000
common_http_port_max     = 8099
common_ip_wait_timeout   = "20m"
common_shutdown_timeout  = "15m"

// HCP Packer
common_hcp_packer_registry_enabled = false

The default provisioning data source for Linux machine image builds is http. This is used to serve the kickstart files to the Linux guest operating system during the build.

common_data_source = "http"

Note

Packer includes a built-in HTTP server that is used to serve the kickstart files for Linux machine image builds.

If iptables/nftables is enabled on your Packer host, you will need to open common_http_port_min through common_http_port_max ports.

iptables command:

iptables -A INPUT -p tcp --match multiport --dports 8000:9000 -j ACCEPT

firewall-cmd example:

firewall-cmd --zone=public --add-port=8000-9000/tcp --permanent
firewall-cmd --reload

You can change the common_data_source from http to disk to build supported Linux machine images without the need to use Packer's HTTP server. This is useful for environments that may not be able to route back to the system from which Packer is running. For example, building a machine image in VMware Cloud on AWS.

common_data_source = "disk"

The Packer plugin's cd_content option is used when selecting disk unless the distribution does not support a secondary CD-ROM.

Name of the network interface that Packer gets HTTPIP from. Defaults to the first non loopback interface.

common_http_interface = "eth2"

IP address on the build server to bind the Packer HTTP instance to. Must be an interface that is reachable from the Proxmox server.

common_http_bind_address = 172.16.15.97"

Edit the config/network.pkrvars.hcl file to configure the following:

These variables are used by Packer to configure the network interface for the VM template. These are specific to your environment. For example, to use the default vmbr0 interface and the tag for VLAN 102, you would set it as follows:

// Proxmox settings for VM templates
vm_bridge_interface  = "vmbr0"
vm_vlan_tag          = "102"

Configuring a static IP address under the configs/network.pkrvars.hcl file is supported. If you want to use DHCP for the templates then leave these variables commented out. The default is DHCP.

Note

• These settings are site specific for each Proxmox host and are going to be needed regardless if you use DHCP or static IP addresses.

The Packer build templates default to using DHCP, however, you can use static IP addressing for your VM templates. Simply uncomment the following vars and configure to your specific requirements:

vm_ip_address = "192.168.101.100"
vm_ip_netmask = 24
vm_ip_gateway = "192.168.101.1"
vm_dns_list   = [ "8.8.8.8", "8.8.4.4" ]

Note

• If you need/want to go back to using DHCP, just comment these variables out again and the templates should go back to using DHCP.

Edit the config/proxmox.pkrvars.hcl file to configure the following:

• Promxox Endpoint and Credentials

// Proxmox Credentials
proxmox_api_token_id        = "name@realm!token"
proxmox_api_token_secret    = "<token secret>"
proxmox_insecure_connection = false

// Proxmox Specific Settings
proxmox_hostname = "<FQDN or IP of proxmox server>"
proxmox_node     = "<proxmox node name>"

The proxmox_api_token_id variable uses a specific format and, as the time of this writing, needs to be assigned to the PVEAdmin role. One of the to-do's is to document a least-privilege method of creating the Proxmox API token.

For more information, please see the Proxmox documentation on authentication.

For Proxmox installs that use a self-signed certificate, you will want to set proxmox_insecure_connection to true.

Edit the config/linux-storage.pkrvars.hcl file to configure storage for VM templates.

// VM Storage Settings
vm_disk_device     = "vda"

vm_disk_device:string - This variable depends on the disk controller used inside of the specific .auto.pkrvars.hcl file. By default, the builds use the virtio-scsi-pci disk controller and that requires the use of vda. If you decide to use a non-virtio controller, then you'll have to change the vm_disk_device variable to the appropriate device.

vm_efi_storage_pool      = "pool0"
vm_efi_type              = "4m"
vm_efi_pre_enrolled_keys = false

vm_disk_partitions:list[dict] - Use this list to define the primary partitions that will be created when a specific build runs. Each of the builds process this list in order, so the first partition defined in the list will be the first partition created, the second one listed will be the second one created, and so on.

Note

• All partition sizes are in MegaBytes (MB)
• If you want to have a partition consume all available free space, you can indicate that with -1

// VM Storage Settings
vm_disk_device     = "vda"
vm_disk_use_swap   = true
vm_disk_partitions = [
  {
    name = "root"
    size = -1,
    format = {
      label  = "ROOTFS",
      fstype = "ext4",
    },
    mount = {
      path    = "/",
      options = "",
    },
    volume_group = "",
  },
]

// VM EFI Settings
vm_efi_storage_pool      = "pool0"
vm_efi_type              = "4m"
vm_efi_pre_enrolled_keys = false

// VM Storage Settings
vm_disk_device     = "vda"
vm_disk_use_swap   = true
vm_disk_partitions = [
  {
    name = "efi"
    size = 1024,
    format = {
      label  = "EFIFS",
      fstype = "fat32",
    },
    mount = {
      path    = "/boot/efi",
      options = "",
    },
    volume_group = "",
  },
  {
    name = "boot"
    size = 1024,
    format = {
      label  = "BOOTFS",
      fstype = "ext4",
    },
    mount = {
      path    = "/boot",
      options = "",
    },
    volume_group = "",
  },
  {
    name = "root"
    size = -1,
    format = {
      label  = "ROOTFS",
      fstype = "ext4",
    },
    mount = {
      path    = "/",
      options = "",
    },
    volume_group = "",
  },
]

//VM EFI Settings
vm_efi_storage_pool      = "pool0"
vm_efi_type              = "4m"
vm_efi_pre_enrolled_keys = false

// UEFI VM Storage Settings
vm_disk_device     = "vda"
vm_disk_use_swap   = true
vm_disk_partitions = [
  {
    name = "efi"
    size = 1024,
    format = {
      label  = "EFIFS",
      fstype = "fat32",
    },
    mount = {
      path    = "/boot/efi",
      options = "",
    },
    volume_group = "",
  },
  {
    name = "boot"
    size = 1024,
    format = {
      label  = "BOOTFS",
      fstype = "ext4",
    },
    mount = {
      path    = "/boot",
      options = "",
    },
    volume_group = "",
  },
  {
    name = "sysvg"
    size = -1,
    format = {
      label  = "",
      fstype = "",
    },
    mount = {
      path    = "",
      options = "",
    },
    volume_group = "sysvg",
  },
]
vm_disk_lvm = [
  {
    name: "sysvg",
    partitions: [
      {
        name = "lv_swap",
        size = 1024,
        format = {
          label  = "SWAPFS",
          fstype = "swap",
        },
        mount = {
          path    = "",
          options = "",
        },
      },
      {
        name = "lv_root",
        size = 10240,
        format = {
          label  = "ROOTFS",
          fstype = "ext4",
        },
        mount = {
          path    = "/",
          options = "",
        },
      },
      {
        name = "lv_home",
        size = 4096,
        format = {
          label  = "HOMEFS",
          fstype = "ext4",
        },
        mount = {
          path    = "/home",
          options = "nodev,nosuid",
        },
      },
      {
        name = "lv_opt",
        size = 2048,
        format = {
          label  = "OPTFS",
          fstype = "ext4",
        },
        mount = {
          path    = "/opt",
          options = "nodev",
        },
      },
      {
        name = "lv_tmp",
        size = 4096,
        format = {
          label  = "TMPFS",
          fstype = "ext4",
        },
        mount = {
          path    = "/tmp",
          options = "nodev,noexec,nosuid",
        },
      },
      {
        name = "lv_var",
        size = 2048,
        format = {
          label  = "VARFS",
          fstype = "ext4",
        },
        mount = {
          path    = "/var",
          options = "nodev",
        },
      },
      {
        name = "lv_var_tmp",
        size = 1000,
        format = {
          label  = "VARTMPFS",
          fstype = "ext4",
        },
        mount = {
          path    = "/var/tmp",
          options = "nodev,noexec,nosuid",
        },
      },
      {
        name = "lv_var_log",
        size = 4096,
        format = {
          label  = "VARLOGFS",
          fstype = "ext4",
        },
        mount = {
          path    = "/var/log",
          options = "nodev,noexec,nosuid",
        },
      },
      {
        name = "lv_var_audit",
        size = 500,
        format = {
          label  = "AUDITFS",
          fstype = "ext4",
        },
        mount = {
          path    = "/var/log/audit",
          options = "nodev,noexec,nosuid",
        },
      },
    ],
  }
]

Edit the <build type>.pkrvars.hcl file in the config directory for each type of build to set the following virtual machine hardware settings, as required:

• CPUs (int)
• CPU Cores (int)
• Memory in MB (int)
• Primary Disk in MB (string) (e.g. 32GB)
• .iso Path (string)
• .iso File (string)

// Guest Operating System Metadata
vm_os_language   = "en_US"
vm_os_keyboard   = "us"
vm_os_timezone   = "UTC"
vm_os_family     = "linux"
vm_os_name       = "ubuntu"
vm_os_version    = "22.04-lts"

// Virtual Machine Guest Operating System Setting
vm_os_type       = "l26"
vm_cloudinit     = true

// Virtual Machine Hardware Settings
vm_bios                 = "ovmf"
vm_cpu_count            = 1
vm_cpu_sockets          = 1
vm_cpu_type             = "kvm64"
vm_mem_size             = 2048
vm_disk_type            = "virtio"
vm_disk_size            = "32G"
vm_disk_format          = "raw"
vm_disk_controller_type = "virtio-scsi-pci"
vm_network_card_model   = "virtio"

// Removable Media Settings
iso_path     = "iso"
iso_file     = "ubuntu-22.04-live-server-amd64.iso"
// The checksum can be a URL or an actual checksum value. URL is preferred
iso_checksum = "file:https://releases.ubuntu.com/jammy/SHA256SUMS"

// Boot Settings
vm_boot      = "order=virtio0;ide2;net0"
vm_boot_wait = "5s"

// EFI Settings
vm_firmware_path         = "./OVMF.fd"

Note

All config/<build type>.pkrvars.hcl default to using:

• VirtIO SCSI storage device
• VirtIO (paravirtualized) network card device
• UEFI boot firmware

The defaults use VirtIO to balance out performance, compatibility, and ease of use. Feel free to change the storage and network controllers to suit your needs. However, if you change the storage or network controllers and run into issues you should change them back to defaults and try the builds again. I won't support any builds that don't use the VirtIO drivers.

Both UEFI and BIOS booting are supported for builds. Inside the <build type>.pkrvars.hcl file specific to the build, you can set the vm_bios variable to either seabios for BIOS or ovmf for UEFI booting.

Note

The storage layouts are different for each bootloader type so you'll need to configure the storage layouts accordingly.

All builds for operating systems that support cloud-init now have the option to enable it. This can be done on a per-build basis inside the <build type>.pkrvars.hcl files in the config directory. The default setting is true.

If a particular linux distribution ships with cloud-init (e.g. Ubuntu) and cloud-init is set to false in the config/<build type>.pkrvars.hcl file for the build, then cloud-init will be disabled in the operating system and within Proxmox for that specific template.

For obvious reasons, product keys and actual ISO names are not provided for any of the Windows builds in this repository. For example, if you wanted to build a Windows 11 template VM, the Windows 11 build uses the Windows 11 Enterprise Eval ISO. By default packer will build the Windows 11 Enterprise Eval whether you choose the Pro or Enterprise build when running build.sh. Also, this build by default should not require any interventions by a human.

If you have a valid product key and want to build a Windows 11 Pro or Enterprise template VM, you will need to change the following variables (if building Pro you don't need to change Enterprise vars and vice versa):

vm_inst_os_image_pro = "Windows 11 Enterprise Evaluation"   <-- If building Professional with a valid key, you will need to enter in whatever the name of the Operating System is when the Windows install prompts you
vm_inst_os_key_pro   = "XXXXX-XXXXX-XXXXX-XXXXX-XXXXX"
vm_inst_os_image_ent = "Windows 11 Enterprise Evaluation"   <-- If building Enterprise with a valid key, you can remove the Evaluation from this string
vm_inst_os_key_ent   = "XXXXX-XXXXX-XXXXX-XXXXX-XXXXX"
...
iso_file             = "Name-of-Windows-11-Non-Evaluation-DVD.iso"  <-- This is the name of the real Windows 11 DVD

If your template build hangs or times out, re-run it and then open the console of the template VM and see if Windows is waiting for you to select an Operating System. Select the Operating System version and then click Next and the build should work hands-off the rest of the way. If you want this to be completely hands-off, just change the vm_inst_os_image_pro and/or vm_inst_os_image_ent vars to match the Operating System selection entry exactly. Then the next time you execute the particular Windows template build, it shouldn't get stuck waiting for input.

Contributions are welcome, please read the CONTRIBUTING document for more details.

The repository is modeled after the VMware Packer Examples repository. As someone who initially struggled with organization of a packer project, the VMware repository helped me significantly.