Bare metal cloud-init

February 18, 2023 by Daniel F Dickinson12 minutes

A guide to using a Debian cloud image and cloud-init on a bare-metal host for fast deployment. The base configuration is very basic and aims to act like an OVH Debian 12 (Bookworm) dedicated server base configuration.

Daniel F. Dickinson’s Cloud-Init deployment for a bare-metal host for a ‘personal infrastructure’. This repository provides the base configuration upon which Ansible will build.

The base configuration is very basic and aims to act like an OVH Debian 12 (Bookworm) dedicated server’s base configuration.

This repository should not include any private or even particularly personalised information in unencrypted form, and is designed for DFD’s uses, so likely not a ‘plug and play’ option for others. It is Daniel’s hope, however, that it still proves informative and useful.

Aside: This method of deployment is much faster at actual deploy time than using a debian-installer based installation. Of course that is because adding packages, setting up users, etc, is work you’ve done ahead of time, and gets applied automagically while you (e.g.) have a coffee. See my old related Ansible repository for an example of the main install.

Metadata

Repository URL

https://gitlab.com/danielfdickinson/debian-bare-metal-cloud-init

Features and default configuration

Sample files for a NoCloud partition for a bare-metal server deployed using a ‘cloud’ image and cloud-init data.
Currently a documentation rather than code project.
- (Daniel’s initial plan was to automate more complex user-data, but that proves unnecessary for the basic configuration that is being imitated).

Using this repository

This repository is meant for the case where you are only rarely deploying on physical hosts (in Daniel’s case the goal is to create local ‘pre-test’ deployments that will be on dedicated servers on OVH for the real deal). If one frequently needs to do this type of deployment, it is instead recommended to set up a proper MaaS (Metal As A Service) environment.

Preparing the ‘cloud-init’ (cloud-config) data

Add your hostname, FQDN, and an SSH public key (for which you have the private key) in both meta-data and user-data
Update the network-config to suit your machine
Optionally, add more disk and filesystem setup to the user-data file
Optionally, unlock the initial deploy user and add a password (if you want to allow console logins immediately, this would be required).

Deploying on the host

Prepare media with the cloud-init data
On a linux-compatible media:
1. Copy meta-data, network-config, and user-data to the media
2. Safely remove the media
Obtain a Debian ‘generic’ cloud image such as https://cloud.debian.org/images/cloud/bookworm/latest/debian-12-generic-amd64.tar.xz
- If you will have a network available when accessing the target host’s storage you may be able to download the image directly onto the desired disk. For example you could do (assuming the host’s root partition belongs on /dev/sdb):
  curl -L https://cloud.debian.org/images/cloud/bookworm/latest/debian-12-generic-amd64.tar.xz | tar -O -xJf - disk.raw | dd of=/dev/sdb bs=1M
- Note that using this technique writes a whole disk image to the target disk, not a single partition or individual files.
- Note 2 If you have access to the storage hardware by itself, you may not needed to boot the system. If the system needs to be booted to access the hardware, a system rescue image such as https://www.system-rescue.org/ can be help. The rescue image can also be handy for other command line system recovery situations).
- An example procedure for this step, when you need to boot the system to access the hardware, might be:
  1. Boot the system using a recovery CD (e.g. System Rescue CD)
  2. Use lsblk -o NAME,SIZE,FSTYPE,LABEL,UUID,MOUNTPOINT to generate a list disks and partitions on the system, e.g.:
    NAME SIZE FSTYPE LABEL UUID MOUNTPOINT sda 931.5G ├─sda1 930.5G ext4 e346a5d8-de78-48eb-c987-67d86b231f10 / ├─sda14 3M └─sda15 124M vfat 3A93-FAF0 /boot/efi sdb 1.7G iso966 RESCUE905 2023-01-31-01-59-58-00 ├─sdb1 730M iso966 RESCUE905 2023-01-31-01-59-58-00 └─sdb2 1.4M vfat FE87-33BA
  3. Identify the rescue disk (in this case sdb) and the target disk (in this case sda).
  4. Configure the network if is not automatically configured. (If you do not have a network at this stage you will need to place the debian cloud image on media such as a USB flash drive).
  5. Since the target disk is /dev/sda (note that we add /dev in front of disk name we identified above) we issue a command such as:
  curl -L https://cloud.debian.org/images/cloud/bookworm/20231218-1609/debian-12-generic-amd64-20231228-1609.tar.xz | tar -O -xJf - disk.raw | dd of=/dev/sda bs=1M
  - Without a network, assuming you copied the debian cloud image to a linux-compatible media, and the media is detected as /dev/sdc, you would instead issue:
    mount /dev/sdc /mnt tar -O -xJf /mnt/debian-12-generic-amd64-20231228-1609.tar.xz disk.raw | dd of=/dev/sda bs=1M umount /mnt
At the end of the boot disk (not partition) add a small FAT32 partition with the filesystem volume label CIDATA. In that partition you should add the files meta-data, user-data, and optional network-config to the root of the filesystem. For more information see the cloud-init 22.4.2 ‘NoCloud documentation.
- (We follow the 22.4.2 documentation because that is the version used with Debian Bookworm cloud images).
- An example procedure that you would use following the steps to extract the debian cloud image, could be:
  1. Execute: parted /dev/sda
  2. See (for example):
    GNU Parted 3.5 Using /dev/sda Welcome to GNU Parted! Type 'help' to view a list of commands.
  3. Enter: print
  4. See:
    Warning: Not all of the space available to /dev/sda appears to be used, you can fix the GPT to use all of the space (an extra 1949330864 blocks) or continue with the current setting? Fix/Ignore?
  5. Type: f to fix the partition table
  6. See (for example):
    Model: ATA BFL38292432 (scsi) Disk /dev/sda: 1000GB Sector size (logical/physical): 512B/512B Partition Table: gpt Disk Flags: Number Start End Size File system Name Flags 14 1049kB 4194kB 3146kB bios_grub 15 4194kB 134MB 130MB fat16 boot, esp 1 134MB 2147MB 2013MB ext4
  7. Enter: mkpart "" fat32 -1G -1
  8. Enter: quit
  9. Format the partition as FAT32 with label CIDATA using:
    mkdosfs -F 32 -n CIDATA /dev/sda2
  10. Insert your media with the cloud-init files
  11. Use lsblk -o NAME,SIZE,FSTYPE,LABEL,UUID,MOUNTPOINT to identify the partition containing the cloud-init files
  12. Assuming /dev/sdc1 has the cloud-init files use:
    mount /dev/sdc1 /mnt
  13. Copy the cloud-init files to the partition you created (e.g. /dev/sda2):
    mkdir /tmp/mnt mount /dev/sda2 /tmp/mnt cp /mnt/{meta-data,network-config,user-data} /tmp/mnt/ umount /tmp/mnt umount /mnt
Reboot to apply the cloud-init config and then login.
1. Ideally, attach a serial connection to the new Debian host
2. On a console that has the other end of the serial connection watch the boot messages after you type:
  reboot
on the new Debian host. 3. cloud-init should set up the system as you defined and now you should be able to SSH into the new Debian host, or access it via serial console or keyboard, video, and mouse console.

Example install (for Debian Bullseye) followed by first boot

GNU GRUB  version 2:2.06.r334.g340377470-1
+----------------------------------------------------------------------------+
|  Boot SystemRescue using default options                                   |
|  Boot SystemRescue and copy system to RAM (copytoram)                      |
| *Boot with serial and copy system to RAM (copytoram)                       |
|  Boot SystemRescue and verify integrity of the medium (checksum)           |
|  [...]                                                                     |
+----------------------------------------------------------------------------+
Use the ^ and v keys to select which entry is highlighted.
Press enter to boot the selected OS, `e' to edit the commands
before booting or `c' for a command-line.

The highlighted entry will be executed automatically in 90s

:: running early hook [udev]
Starting version 251.6-2-arch

[...]

:: Mounting '/dev/disk/by-label/RESCUE905' to '/run/archiso/bootmnt'
:: Device '/dev/disk/by-label/RESCUE905' mounted successfully.

[...]

Welcome to SystemRescue 9.05!

[  OK  ] Created slice Slice /system/getty.
[  OK  ] Created slice Slice /system/modprobe.
[...]
Starting D-Bus System Message Bus...
Starting SSH Key Generation...
Starting SystemRescue Initialization, before networking...
Starting User Login Management...
[  OK  ] Started D-Bus System Message Bus.
[  OK  ] Started User Login Management.
[  OK  ] Listening on Load/Save RF &hellip;itch Status /dev/rfkill Watch.
[...]
[  OK  ] Reached target Preparation for Logins.
[  OK  ] Started Getty on tty1.
[  OK  ] Started Serial Getty on ttyS0.
[  OK  ] Reached target Login Prompts.
[  OK  ] Reached target Multi-User System.

========= SystemRescue 9.05 (x86_64) ======== ttyS0/6 =========
https://www.system-rescue.org/

* Console environment :
  Run setkmap to choose the keyboard layout (also accessible with the arrow up key)
  Run manual to read the documentation of SystemRescue

* Graphical environment :
  Type startx to run the graphical environment
  X.Org comes with the XFCE environment and several graphical tools:
  - Partition manager: .. gparted
  - Web browser: ........ firefox
  - Text editor: ........ featherpad

sysrescue login: root (automatic login)

[root@sysrescue ~]# lsblk -o NAME,SIZE,FSTYPE,LABEL,UUID,MOUNTPOINT
NAME   SIZE FSTYPE LABEL     UUID              MOUNTPOINT
loop0  656.6M squash                          /run/archi
sda    931.5G
├─sda1 930.5G ext4           e9d180ca-4d05-4e4f-b99e-8c49d6ff5ee7
├─sda2 953M vfat   CIDATA    0E6B-8C93
├─sda14 3M
└─sda15 124M vfat            D2A8-932D
sdb    1.8T
└─sdb1 1.8T LVM2_m           etC2dD-f1ph-2Ilc-s9IS-0b9f-k3Zf-D40IOW
sdc    1.8T
└─sdc1 1.8T LVM2_m           fvu0Ic-eyX1-3Tpp-02k4-B4zv-tImL-bZ01et
sdd    0B
sde    0B
sdf    0B
sdg    0B
sdh    1.7G iso966 RESCUE905 2023-01-31-01-59-58-00
├─sdh1 730M iso966 RESCUE905 2023-01-31-01-59-58-00
└─sdh2 1.4M vfat             7F86-83CD
sr0    1024M

[root@sysrescue ~]# eject /dev/sdh
[root@sysrescue ~]# lsblk -oe NAME,SIZE,FSTYPE,LABEL,UUID,MOUNTPOINT
NAME   SIZE FSTYPE LABEL     UUID              MOUNTPOINT
loop0  656.6M squash                          /run/archi
sda    931.5G
├─sda1 930.5G ext4           e9d180ca-4d05-4e4f-b99e-8c49d6ff5ee7
├─sda2 953M vfat   CIDATA    0E6B-8C93
├─sda14 3M
└─sda15 124M vfat            D2A8-932D
sdb    1.8T
└─sdb1 1.8T LVM2_m           etC2dD-f1ph-2Ilc-s9IS-0b9f-k3Zf-D40IOW
sdc    1.8T
└─sdc1 1.8T LVM2_m           fvu0Ic-eyX1-3Tpp-02k4-B4zv-tImL-bZ01et
sdd    0B
sde    0B
sdf    0B
sdg    0B
sr0    1024M
[root@sysrescue ~]# lsblk -o NAME,SIZE,FSTYPE,LABEL,UUID,MOUNTPOINT
NAME   SIZE FSTYPE LABEL     UUID              MOUNTPOINT
loop0  656.6M squash                          /run/archi
sda    931.5G
├─sda1 930.5G ext4           e9d180ca-4d05-4e4f-b99e-8c49d6ff5ee7
├─sda2 953M vfat   CIDATA    0E6B-8C93
├─sda14 3M
└─sda15 124M vfat            D2A8-932D
sdb    1.8T
└─sdb1 1.8T LVM2_m           etC2dD-f1ph-2Ilc-s9IS-0b9f-k3Zf-D40IOW
sdc    1.8T
└─sdc1 1.8T LVM2_m           fvu0Ic-eyX1-3Tpp-02k4-B4zv-tImL-bZ01et
sdd    0B
sde    0B
sdf    0B
sdg    0B
sr0    1024M
sdh    7.3G ext4   Bare-metal via c   xxxxxxxxxxx-xxxxxx-xxxxx-xxxxx
sr0   1024M
[root@sysrescue ~]# mount /dev/sdh /mnt
[root@sysrescue ~]# cd /mnt
[root@sysrescue /mnt]# ls
debian-11-generic-amd64-20230124-1270.tar.xz  lost+found  nocloud
[root@sysrescue /mnt]# mkdir /media
[root@sysrescue /mnt]# tar -O -xvJf debian-11-generic-amd64-20230124-1270.tar.xz disk.raw | dd of=/dev/sda1 bs=1M
disk.raw
0+190212 records in
0+190212 records out
2147483648 bytes (2.1 GB, 2.0 GiB) copied, 30.1568 s, 71.2 MB/s
[root@sysrescue /mnt]# parted /dev/sda
GNU Parted 3.5
Using /dev/sda
Welcome to GNU Parted! Type 'help' to view a list of commands.
(parted) print
Warning: Not all of the space available to /dev/sda appears to be used, you can
fix the GPT to use all of the space (an extra 1949330864 blocks) or continue
with the current setting?
Fix/Ignore? fix
Model: ATA CT1000BX100SSD1 (scsi)
Disk /dev/sda: 1000GB
Sector size (logical/physical): 512B/512B
Partition Table: gpt
Disk Flags:

Number Start    End       Size      File system  Name  Flags
14     1049kB   4194kB    3146kB    bios_grub
15     4194kB    134MB    130MB     fat16              boot, esp
1       134MB   2147MB   2013MB     ext4

(parted) mkpart "" fat32 -1G -1
(parted) print
Model: ATA CT1000BX100SSD1 (scsi)
Disk /dev/sda: 1000GB
Sector size (logical/physical): 512B/512B
Partition Table: gpt
Disk Flags:

Number Start    End       Size      File system  Name  Flags
14     1049kB   4194kB    3146kB    bios_grub
15     4194kB    134MB    130MB     fat16              boot, esp
1       134MB   2147MB   2013MB     ext4
2       999GB   1000GB    999MB     fat32              msftdata

(parted) quit
Information: You may need to update /etc/fstab.

[root@sysrescue /mnt]# mkfs.vfat -F 32 -n CIDATA /dev/sda2
mkfs.fat 4.2 (2021-01-31)
[root@sysrescue /mnt]# mount /dev/sda2 /media
[root@sysrescue /mnt]# ls /media/
[root@sysrescue /mnt]# mount
proc on /proc type proc (rw,nosuid,nodev,noexec,relatime)
sys on /sys type sysfs (rw,nosuid,nodev,noexec,relatime)
[...]
tmpfs on /run/user/0 type tmpfs (rw,nosuid,nodev,relatime,size=3180068k,nr_inodes=795017,mode=700,inode64)
/dev/sdh on /mnt type ext4 (rw,relatime)
/dev/sda2 on /media type vfat (rw,relatime,fmask=0022,dmask=0022,codepage=437,iocharset=ascii,shortname=mixed,utf8,errors=remount-ro)
[root@sysrescue /mnt]# ls nocloud/
meta-data  network-config  user-data
[root@sysrescue /mnt]# cp nocloud/* /media/
[root@sysrescue /mnt]# ls /media/
meta-data  network-config  user-data
[root@sysrescue /mnt]# umount /media
[root@sysrescue /mnt]# cd /
[root@sysrescue /mnt]# eject /dev/sdh
[root@sysrescue /]# mount
proc on /proc type proc (rw,nosuid,nodev,noexec,relatime)
sys on /sys type sysfs (rw,nosuid,nodev,noexec,relatime)
[...]
tmpfs on /run/user/0 type tmpfs (rw,nosuid,nodev,relatime,size=3180068k,nr_inodes=795017,mode=700,inode64)

tar -O -xvJf debian-11-generic-amd64-20230124-1270.tar.xz disk.raw | dd of=/dev/mkdirs/media
lsblk -o NAME,SIZE,FSTYPE,LABEL,UUID,MOUNTPOINT
NAME    SIZE    FSTYPE  LABEL         UUID             MOUNTPOINT
loop0    656.6M squashf                               /run/archis
sda      931.5G
├─sda1     1.9G ext4                    e9d180ca-4d05-4e4f-b99e-8c49d6ff5ee7
├─sda2     953M vfat    CIDATA        9B59-C16F
├─sda14      3M
└─sda15    124M vfat                  D2A8-932D
sdb        1.8T
└─sdb1     1.8T LVM2_me               ftC6dA-fOpg-1Ikc-s0YS-0b9f-k3Zf-D40IOW
sdc        1.8T
└─sdc1     1.8T LVM2_me               fbu0Pc-esXk-3Too-01t5-B4zv-tImL-bZ01et
sdd          0B
sde          0B
sdf          0B
sdg          0B
sr0       1024M
[root@sysrescue /]# reboot
[root@sysrescue /]#
  Stopping Session 1 of User root...
  Stopping Session 2 of User root...
[  OK  ] Removed slice Slice /system/modprobe.
[  OK  ] Stopped target Multi-User System.
[...]
[  OK  ] Reached target System Shutdown.
[  OK  ] Reached target Late Shutdown Services.
[  OK  ] Finished System Reboot.
[  OK  ] Reached target System Reboot.
[...]
[  393.223783] shutdown[1]: Failed to finalize file systems, loop devices, ignoring.
[  393.508673] reboot: Restarting system
--------------------------------------------------------------------------------
[    0.000000] Linux version 5.10.0-21-amd64 (debian-kernel@lists.debian.org) (gcc-10 (Debian 10.2.1-6) 10.2.1 20210110, GNU ld (GNU Binutils for Debian) 2.35.2) #1 SMP Debian 5.10.162-1 (2023-01-21)
[    0.000000] Command line: BOOT_IMAGE=/boot/vmlinuz-5.10.0-21-amd64 root=UUID=xxxxxx-xxxxxx-xxxxxx-xxxx ro console=tty0 console=ttyS0,115200 earlyprintk=ttyS0,115200 consoleblank=0
[    0.000000] x86/fpu: Supporting XSAVE feature 0x001: 'x87 floating point registers'
[    0.000000] x86/fpu: Supporting XSAVE feature 0x002: 'SSE registers'
[    0.000000] x86/fpu: Supporting XSAVE feature 0x004: 'AVX registers'
[    0.000000] x86/fpu: xstate_offset[2]:  576, xstate_sizes[2]:  256
[    0.000000] x86/fpu: Enabled xstate features 0x7, context size is 832 bytes, using 'standard' format.
[    0.000000] BIOS-provided physical RAM map:
[...]
[    5.499258] usbcore: registered new interface driver uas
[    5.503608] hub 8-0:1.0: 5 ports detected
Begin: Loading essential drivers ... done.
[...]
Begin: Running /scripts/init-premount ... done.
[...]
Welcome to Debian GNU/Linux 11 (bullseye)!

[    6.642082] systemd[1]: Set hostname to <debian>.
[    6.648738] random: systemd: uninitialized urandom read (16 bytes read)
[...]
[  OK  ] Finished Load AppArmor profiles.
[  OK  ] Started ifup for enp1s0.
[  OK  ] Started ifup for enp3s0.
  Starting Raise network interfaces...
[*     ] A start job is running for Raise network interfaces (5s / 5min 3s)
[   12.698722] e1000e 0000:03:00.0 enp3s0: NIC Link is Up 1000 Mbps Full Duplex, Flow Control: Rx/Tx
[   12.709838] IPv6: ADDRCONF(NETDEV_CHANGE): enp3s0: link becomes ready
[   12.782696] e1000e 0000:01:00.0 enp1s0: NIC Link is Up 1000 Mbps Full Duplex, Flow Control: Rx/Tx
[   12.794432] IPv6: ADDRCONF(NETDEV_CHANGE): enp1s0: link becomes ready
[...]
[  OK  ] Finished Raise network interfaces.
[  OK  ] Reached target Network.
  Starting Initial cloud-init (metadata service crawler)...
[   20.902510] cloud-init[727]: Cloud-init v. 20.4.1 running 'init' at Tue, 31 Jan 2023 14:49:11 +0000. Up 20.87 seconds.
[   20.924705] cloud-init[727]: ci-info: +++++++++++++++++++++++++++++++++++++++Net device info+++++++++++++++++++++++++++++++++++++++
[   20.951306] cloud-init[727]: ci-info: +--------+------+------------------------------+--------------+--------+-------------------+
[   20.975323] cloud-init[727]: ci-info: | Device |  Up  |     Address                  |      Mask    | Scope  |   Hw-Address      |
[   20.999314] cloud-init[727]: ci-info: +--------+------+------------------------------+--------------+--------+-------------------+
[...]
[   23.502767] cloud-init[808]: Cloud-init v. 20.4.1 running 'modules:config' at Tue, 31 Jan 2023 14:49:13 +0000. Up 22.35 seconds.
[   23.528529] cloud-init[824]: Cloud-init v. 20.4.1 running 'modules:final' at Tue, 31 Jan 2023 14:49:14 +0000. Up 23.34 seconds.
[   23.565264] cloud-init[824]: Get:1 http://security.debian.org/debian-security bullseye-security InRelease [48.4 kB]
[   23.583425] cloud-init[824]: Get:2 http://deb.debian.org/debian bullseye InRelease [116 kB]
[   23.626995] cloud-init[824]: Get:3 http://deb.debian.org/debian bullseye-updates InRelease [44.1 kB]
[...]

Debian GNU/Linux 11 dlcicl01 ttyS0

dlcicl01 login:

[   28.875554] cloud-init[824]: Fetched 25.0 MB in 5s (4629 kB/s)
[   29.773719] cloud-init[824]: Reading package lists...
[   29.858709] cloud-init[824]: Reading package lists...
[   30.074568] cloud-init[824]: Building dependency tree...
[   30.074802] cloud-init[824]: Reading state information...
[   30.339086] cloud-init[824]: Calculating upgrade...
[   30.629502] cloud-init[824]: The following packages will be upgraded:
[   30.632259] cloud-init[824]:     bind9-host bind9-libs curl libcurl3-gnutls libcurl4
[   30.747156] cloud-init[824]: 5 upgraded, 0 newly installed, 0 to remove and 0 not upgraded.
[...]
[   33.768063] cloud-init[824]: Cloud-init v. 20.4.1 finished at Tue, 31 Jan 2023 14:49:24 +0000. Datasource DataSourceNoCloud [seed=/dev/sda2][dsmode=net].  Up 33.76 seconds

dlcicl01 login:

Wild Tech Garden

Title here

Bare metal cloud-init

Metadata

Repository URL

Features and default configuration

Using this repository

Preparing the ‘cloud-init’ (cloud-config) data

Deploying on the host

Bare metal cloud-init

Metadata

Repository URL

Features and default configuration

Using this repository

Preparing the ‘cloud-init’ (cloud-config) data

Deploying on the host

Related posts

Pi Unbound: The gift of DoT

UEFI automated ARM for Libvirt/KVM

Old school automated ARM for Libvirt/KVM