Files
oam/knowledge base/cloud-init.md
2023-01-14 12:11:10 +01:00

4.7 KiB

Cloud init

  1. TL;DR
  2. Merge 2 or more files or parts
    1. In Terraform
  3. Further readings
  4. Sources

TL;DR

# Get the current status.
cloud-init status
cloud-init status --wait

# Verify that cloud-init received the expected user data.
cloud-init query userdata

# Assert the user data we provided is a valid cloud-config.
# From version 22.2, drops the 'devel' command.
cloud-init devel schema --system --annotate
cloud-init devel schema --config-file '/tmp/user-data'

# Check the raw logs.
cat '/var/log/cloud-init.log'
cat '/var/log/cloud-init-output.log'

# Parse and organize the events in the log file by stage.
cloud-init analyze show

# Manually run a single cloud-config module once after the instance has booted.
sudo cloud-init single --name 'cc_ssh' --frequency 'always'

# Clean up everything so `cloud-init` can run again.
sudo cloud-init clean

# Re-run everything.
sudo cloud-init init

# Check the user scripts.
ls '/var/lib/cloud/instance/scripts'
#cloud-config

# Add the Docker repository
yum_repos:
  docker-ce:
    name: Docker CE Stable - $basearch
    enabled: true
    baseurl: https://download.docker.com/linux/rhel/$releasever/$basearch/stable
    priority: 1
    gpgcheck: true
    gpgkey: https://download.docker.com/linux/rhel/gpg

# Upgrade the instance
package_upgrade: true
package_reboot_if_required: true

# Install required packages
# docker-ce already depends on docker-ce-cli and containerd.io
packages:
  - docker-ce
  - jq
  - unzip

# Enable and start the service after installation
runcmd:
  - systemctl daemon-reload
  - systemctl enable --now docker.service

Merge 2 or more files or parts

FIXME

See Merging User-Data sections for details.

#cloud-config
packages:
  - jq
  - unzip

---
merge_how:
 - name: list
   settings: [append]
 - name: dict
   settings: [no_replace, recurse_list]

packages:
  - parallel

---
packages:
  - vim

merge_type: 'list(append)+dict(recurse_array)+str()'

In Terraform

  1. create a data resource containing the files in order, one per part:

    # https://registry.terraform.io/providers/hashicorp/cloudinit/latest/docs
    # https://github.com/chrusty/terraform-multipart-userdata/blob/master/example/cloudinit.tf
    data "cloudinit_config" "vm" {
      gzip = true
      base64_encode = true
    
      part {
        content      = file("files/first.yaml")
        content_type = "text/cloud-config"
      }
      
      part {
        content      = file("files/n-th.yaml")
        content_type = "text/cloud-config"
        filename     = "n-th.yaml"
        merge_type   = "dict(recurse_array,no_replace)+list(append)"
      }
    }
    
  2. give its rendered form as input to a vm's userdata attribute or an output resource:

    resource "azurerm_linux_virtual_machine" "vm" {
      user_data = data.cloudinit_config.vm.rendered
      
    }
    
    output "cloudinit_config" {
      value = data.cloudinit_config.vm.rendered
    }
    

Further readings

Sources