Harvester Configuration

Configuration Example

Harvester configuration file can be provided during manual or automatic installation to configure various settings. The following is a configuration example:

  1. scheme_version: 1
  2. server_url: https://cluster-VIP:443
  3. token: TOKEN_VALUE
  4. os:
  5. ssh_authorized_keys:
  6. - ssh-rsa AAAAB3NzaC1yc2EAAAADAQAB...
  7. - github:username
  8. write_files:
  9. - encoding: ""
  10. content: test content
  11. owner: root
  12. path: /etc/test.txt
  13. permissions: '0755'
  14. hostname: myhost
  15. modules:
  16. - kvm
  17. - nvme
  18. sysctls:
  19. kernel.printk: "4 4 1 7"
  20. kernel.kptr_restrict: "1"
  21. dns_nameservers:
  22. - 8.8.8.8
  23. - 1.1.1.1
  24. ntp_servers:
  25. - 0.suse.pool.ntp.org
  26. - 1.suse.pool.ntp.org
  27. password: rancher
  28. environment:
  29. http_proxy: http://myserver
  30. https_proxy: http://myserver
  31. labels:
  32. topology.kubernetes.io/zone: zone1
  33. foo: bar
  34. mylabel: myvalue
  35. install:
  36. mode: create
  37. management_interface:
  38. interfaces:
  39. - name: ens5
  40. hwAddr: "B8:CA:3A:6A:64:7C"
  41. method: dhcp
  42. force_efi: true
  43. device: /dev/sda
  44. data_disk: /dev/sdb
  45. silent: true
  46. iso_url: http://myserver/test.iso
  47. poweroff: true
  48. no_format: true
  49. debug: true
  50. tty: ttyS0
  51. vip: 10.10.0.19
  52. vip_hw_addr: 52:54:00:ec:0e:0b
  53. vip_mode: dhcp
  54. force_mbr: false
  55. addons:
  56. harvester_vm_import_controller:
  57. enabled: false
  58. values_content: ""
  59. harvester_pcidevices_controller:
  60. enabled: false
  61. values_content: ""
  62. rancher_monitoring:
  63. enabled: true
  64. values_content: ""
  65. rancher_logging:
  66. enabled: false
  67. values_content: ""
  68. harvester_seeder:
  69. enabled: false
  70. values_content: ""
  71. system_settings:
  72. auto-disk-provision-paths: ""

Configuration Reference

Below is a reference of all configuration keys.

Harvester Configuration - 图1caution

Security Risks: The configuration file contains credentials which should be kept secret. Please do not make the configuration file publicly accessible.

Harvester Configuration - 图2note

Configuration Priority: When you provide a remote Harvester Configuration file during the install of Harvester, the Harvester Configuration file will not overwrite the values for the inputs you had previously filled out and selected. Priority is given to the values that you input during the guided install. For instance, if you have in your Harvester Configuration file specified os.hostname and during install you fill in the field of hostname when prompted, the value that you filled in will take priority over your Harvester Configuration’s os.hostname.

scheme_version

Definition

The version of scheme reserved for future configuration migration.

This configuration is mandatory for migrating the configuration to a new scheme version. It tells Harvester the previous version and the need to migrate.

Harvester Configuration - 图3note

This field didn’t take any effect in the current Harvester version.

Harvester Configuration - 图4caution

Make sure that your custom configuration always has the correct scheme version.

server_url

Definition

server_url is the URL of the Harvester cluster, which is used for the new node to join the cluster.

This configuration is mandatory when the installation is in JOIN mode. The default format of server_url is https://cluster-VIP:443.

Harvester Configuration - 图5note

To ensure a high availability (HA) Harvester cluster, please use either the Harvester cluster VIP or a domain name in server_url.

Example

  1. server_url: https://cluster-VIP:443
  2. install:
  3. mode: join

token

Definition

The cluster secret or node token. If the value matches the format of a node token it will automatically be assumed to be a node token. Otherwise it is treated as a cluster secret.

In order for a new node to join the Harvester cluster, the token should match what the server has.

Example

  1. token: myclustersecret

Or a node token

  1. token: "K1074ec55daebdf54ef48294b0ddf0ce1c3cb64ee7e3d0b9ec79fbc7baf1f7ddac6::node:77689533d0140c7019416603a05275d4"

os.ssh_authorized_keys

Definition

A list of SSH authorized keys that should be added to the default user, rancher. SSH keys can be obtained from GitHub user accounts by using the format github:${USERNAME}. This is done by downloading the keys from https://github.com/${USERNAME}.keys.

Example

  1. os:
  2. ssh_authorized_keys:
  3. - "ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAABAQC2TBZGjE+J8ag11dzkFT58J3XPONrDVmalCNrKxsfADfyy0eqdZrG8hcAxAR/5zuj90Gin2uBR4Sw6Cn4VHsPZcFpXyQCjK1QDADj+WcuhpXOIOY3AB0LZBly9NI0ll+8lo3QtEaoyRLtrMBhQ6Mooy2M3MTG4JNwU9o3yInuqZWf9PvtW6KxMl+ygg1xZkljhemGZ9k0wSrjqif+8usNbzVlCOVQmZwZA+BZxbdcLNwkg7zWJSXzDIXyqM6iWPGXQDEbWLq3+HR1qKucTCSxjbqoe0FD5xcW7NHIME5XKX84yH92n6yn+rxSsyUfhJWYqJd+i0fKf5UbN6qLrtd/D"
  4. - "github:ibuildthecloud"

os.write_files

A list of files to write to disk on boot. The encoding field specifies the content’s encoding. Valid encoding values are:

  • "": content data are written in plain text. In this case, the encoding field can be also omitted.
  • b64, base64: content data are base64-encoded.
  • gz, gzip: content data are gzip-compressed.
  • gz+base64, gzip+base64, gz+b64, gzip+b64: content data are gzip-compressed first and then base64-encoded.

Example

  1. os:
  2. write_files:
  3. - encoding: b64
  4. content: CiMgVGhpcyBmaWxlIGNvbnRyb2xzIHRoZSBzdGF0ZSBvZiBTRUxpbnV4...
  5. owner: root:root
  6. path: /etc/connman/main.conf
  7. permissions: '0644'
  8. - content: |
  9. # My new /etc/sysconfig/samba file
  10. SMDBOPTIONS="-D"
  11. path: /etc/sysconfig/samba
  12. - content: !!binary |
  13. f0VMRgIBAQAAAAAAAAAAAAIAPgABAAAAwARAAAAAAABAAAAAAAAAAJAVAAAAAA
  14. AEAAHgAdAAYAAAAFAAAAQAAAAAAAAABAAEAAAAAAAEAAQAAAAAAAwAEAAAAAAA
  15. AAAAAAAAAwAAAAQAAAAAAgAAAAAAAAACQAAAAAAAAAJAAAAAAAAcAAAAAAAAAB
  16. ...
  17. path: /bin/arch
  18. permissions: '0555'
  19. - content: |
  20. 15 * * * * root ship_logs
  21. path: /etc/crontab

os.persistent_state_paths

Definition

The os.persistent_state_paths option allows you to configure custom paths where modifications made to files will persist across reboots. Any changes to files in these paths will not be lost after a reboot.

Example

Refer to the following example config for installing rook-ceph in Harvester:

  1. os:
  2. persistent_state_paths:
  3. - /var/lib/rook
  4. - /var/lib/ceph
  5. modules:
  6. - rbd
  7. - nbd

os.after_install_chroot_commands

Definition

You can add additional software packages with after_install_chroot_commands. The after-install-chroot stage, provided by elemental-toolkit, allows you to execute commands not restricted by file system write issues, ensuring the persistence of user-defined commands even after a system reboot.

Example

Refer to the following example config for installing an RPM package in Harvester:

  1. os:
  2. after_install_chroot_commands:
  3. - rpm -ivh <the url of rpm package>

DNS resolution is unavailable in the after-install-chroot stage, and the nameserver might not be available. If you need to access a domain name to install a package using an URL, create a temporary /etc/resolv.conf file first. For example:

  1. os:
  2. after_install_chroot_commands:
  3. - "rm -f /etc/resolv.conf && echo 'nameserver 8.8.8.8' | sudo tee /etc/resolv.conf"
  4. - "mkdir /usr/local/bin"
  5. - "curl -fsSL -o get_helm.sh https://raw.githubusercontent.com/helm/helm/main/scripts/get-helm-3 && chmod 700 get_helm.sh && ./get_helm.sh"
  6. - "rm -f /etc/resolv.conf && ln -s /var/run/netconfig/resolv.conf /etc/resolv.conf"

Harvester Configuration - 图6note

Upgrading Harvester causes the changes to the OS in the after-install-chroot stage to be lost. You must also configure the after-upgrade-chroot to make your changes persistent across an upgrade. Refer to Runtime persistent changes before upgrading Harvester.

os.hostname

Definition

Set the system hostname. The installer will generate a random hostname if the user doesn’t provide a value.

Example

  1. os:
  2. hostname: myhostname

os.modules

Definition

A list of kernel modules to be loaded on start.

Example

  1. os:
  2. modules:
  3. - kvm
  4. - nvme

os.sysctls

Definition

Kernel sysctl to set up on start. These are the typical configurations found in /etc/sysctl.conf. Values must be specified as strings.

Example

  1. os:
  2. sysctls:
  3. kernel.printk: 4 4 1 7 # the YAML parser will read as a string
  4. kernel.kptr_restrict: "1" # force the YAML parser to read as a string

os.dns_nameservers

Definition

Fallback DNS name servers to use if DNS is not configured by DHCP or in the OS.

Example

  1. os:
  2. dns_nameservers:
  3. - 8.8.8.8
  4. - 1.1.1.1

os.ntp_servers

Definition

Fallback ntp servers to use if NTP is not configured elsewhere in the OS. Highly recommend to configure os.ntp_servers to avoid time synchronization issue between machines.

Example

  1. os:
  2. ntp_servers:
  3. - 0.suse.pool.ntp.org
  4. - 1.suse.pool.ntp.org

os.password

Definition

The password for the default user, rancher. By default, there is no password for the rancher user. If you set a password at runtime it will be reset on the next boot. The value of the password can be clear text or an encrypted form. The easiest way to get this encrypted form is to change your password on a Linux system and copy the value of the second field from /etc/shadow. You can also encrypt a password using OpenSSL. For the encryption algorithms supported by Harvester, please refer to the table below.

AlgorithmCommandSupport
SHA-512openssl passwd -6Yes
SHA-256openssl passwd -5Yes
MD5openssl passwd -1Yes
MD5, Apache variantopenssl passwd -apr1Yes
AIX-MD5openssl passwd -aixmd5No

Example

Encrypted:

  1. os:
  2. password: "$6$kZYUnRaTxNdg4W8H$WSEJydGWsNpaRbbbRdTDLJ2hDLbkizxSFGW2RtexlqG6njEATaGQG9ssztjaKDCsaNUPBZ1E1YdsvSLMAi/IO/"

Or clear text:

  1. os:
  2. password: supersecure

os.environment

Definition

Environment variables to be set on K3s and other processes like the boot process. Primary use of this field is to set the HTTP proxy.

Example

  1. os:
  2. environment:
  3. http_proxy: http://myserver
  4. https_proxy: http://myserver

Harvester Configuration - 图7note

This example sets the HTTP(S) proxy for foundational OS components. To set up an HTTP(S) proxy for Harvester components such as fetching external images and backup to S3 services, see Settings/http-proxy.

os.labels

Definition

Labels to be added to this Node.

Example

  1. os:
  2. labels:
  3. topology.kubernetes.io/zone: zone1
  4. foo: bar
  5. mylabel: myvalue

os.sshd.sftp

Definition

Subsystem used to configure the OpenSSH Daemon (sshd). Harvester currently only supports sftp.

Example

  1. os:
  2. sshd:
  3. sftp: true # The SFTP subsystem is enabled.

install.addons

Versions: v1.2.0 and later

Definition: Setting that defines the default addon status. Harvester addons are disabled by default.

Supported values:

  • vm-import-controller (chartName: harvester-vm-import-controller)
  • pcidevices-controller (chartName: harvester-pcidevices-controller)
  • rancher-monitoring
  • rancher-logging
  • harvester-seeder (experimental)

Example:

  1. install:
  2. addons:
  3. rancher_monitoring:
  4. enabled: true
  5. rancher_logging:
  6. enabled: false

install.automatic

Definition: Setting that forces the installer to skip the interactive steps in the installation process.

When enabled, the configuration is either retrieved from the value of harvester.install.config_url or defined individually using kernel parameters.

install.data_disk

Versions: v1.0.1 and later

Definition: Default device for storing VM data.

When installing via PXE, use /dev/disk/by-id/$id or /dev/disk/by-path/$path to specify the storage device if the server contains multiple physical volumes.

Default value: Storage device configured in the setting install.device

Example:

  1. install:
  2. data_disk: /dev/sdb

install.debug

Definition: Setting that enables additional logging and debugging during installation.

install.device

Definition: Device on which the Harvester OS is installed.

When installing via PXE, use /dev/disk/by-id/$id or /dev/disk/by-path/$path to specify the storage device if the server contains multiple physical volumes.

install.force_efi

Definition: Setting that forces EFI installation even when EFI is not detected.

Default value: false

install.force_mbr

Definition: Setting that forces usage of MBR partitioning on BIOS systems.

Harvester uses GPT partitioning on UEFI and BIOS systems by default. Compatibility issues may require you to use MBR partitioning instead.

If you specify the same storage device for both install.device and install.data_disk, Harvester creates an additional partition for storing VM data. This additional partition is not created when you force usage of MBR partitioning. Instead, VM data is stored in a partition that stores OS data.

Example:

  1. install:
  2. force_mbr: true

install.harvester.longhorn.default_settings.guaranteedInstanceManagerCPU

Versions: v1.4.0 and later

Definition: Percentage of the total allocatable CPU on each node to be reserved for each Longhorn Instance Manager pod.

Using the default value is recommended for high system availability. When deploying single-node Harvester clusters, you can specify a value less than 12.

For more information about how to set the correct value, see Guaranteed Instance Manager CPU in the Longhorn documentation.

Default value: 12

Supported values: 0 to 12. All other values are considered 12.

Example:

  1. harvester:
  2. longhorn:
  3. default_settings:
  4. guaranteedInstanceManagerCPU: 6

install.harvester.storage_class.replica_count

Versions: v1.1.2 and later

Definition: Replica count of the default Harvester StorageClass harvester-longhorn.

Using the default value is recommended for high storage availability. When deploying single-node Harvester clusters, you must set the value to 1.

For more information, see Default Replica Count in the Longhorn documentation.

Default value: 3

Supported values: 1 to 3. All other values are considered 3.

Example:

  1. install:
  2. harvester:
  3. storage_class:
  4. replica_count: 1

install.iso_url

Definition: URL of ISO image to be downloaded and used to install Harvester when booting from the kernel or vmlinuz.

install.management_interface

Definition: Network interfaces for the host machine.

Harvester uses the systemd net naming scheme. Ensure that the interface name is present on the target machine before installation.

Fields:

  • method: Method used to assign an IP to the network. Supported values:
    • dhcp: Harvester requests an IP from the DHCP server.
    • static: IP and gateway addresses are manually assigned.
  • ip: Static IP assigned to the network. This field is required when the value of method is static.
  • subnet_mask: Subnet mask of the network. This field is required when the value of method is static.
  • gateway: Gateway address assigned to the network. This field is required when the value of method is static.
  • interfaces: Array of network interfaces. The installer combines the specified interfaces (slaves) into a single logical bonded interface.
    • interfaces.name: Name of a slave interface.
    • interfaces.hwAddr: Hardware MAC address of a slave interface. This field is optional.
  • bond_options: Options for bonded interfaces. When unspecified, the following options are used:
    • mode: balance-tlb
    • miimon: 100
  • mtu: Maximum transmission unit (MTU) for the interface.
  • vlan_id: VLAN ID for the interface.

Example:

  1. install:
  2. mode: create
  3. management_interface:
  4. interfaces:
  5. - name: ens5
  6. hwAddr: "B8:CA:3A:6A:64:7D" # Optional
  7. method: dhcp
  8. bond_options:
  9. mode: balance-tlb
  10. miimon: 100
  11. mtu: 1492
  12. vlan_id: 101

install.mode

Definition: Mode of installing Harvester.

Supported values:

  • create: Create a new Harvester installation.
  • join: Join an existing Harvester installation. You must specify the server_url.

Example:

  1. install:
  2. mode: create

install.no_format

Definition: Setting that prevents partitioning and formatting of the installation disk.

install.persistent_partition_size

Definition: Size of the partition COS_PERSISTENT in Gi or Mi.

This partition stores data such as system packages and container images. The minimum value is 150 Gi.

Default value: 150 Gi

Example:

  1. install:
  2. persistent_partition_size: 150Gi

install.poweroff

Definition: Setting that shuts down (instead of rebooting) the server after installation.

install.rawdiskimagepath

Definition: Setting that forces the installer to only install the Harvester hypervisor (without any configuration). You must enable harvester.install.automatic to use this setting.

install.role

Definition: Role assigned to a node at the time of installation. When unspecified, Harvester assigns the default role.

  • default: Allows a node to function as a management node or a worker node.
  • management: Allows a node to be prioritized when Harvester promotes nodes to management nodes.
  • worker: Restricts a node to being a worker node (never promoted to management node) in a specific cluster.
  • witness: Restricts a node to being a witness node (only functions as an etcd node) in a specific cluster.

install.silent

Definition: Reserved

install.skipchecks

Definition: Setting that allows installation to proceed even if minimum requirements for production use are not met

The installer automatically checks if the hardware meets the minimum requirements for production use. When performing automated installation via PXE Boot, if any of the checks fail, installation is stopped, and warnings are printed to the system console and saved to /var/log/console.log in the installation environment.

To override this behavior, set install.skipchecks=true. When set to true, warning messages are still saved to /var/log/console.log, but the installation proceeds even if hardware requirements for production use are not met.

Default value: false

Example:

  1. install:
  2. skipchecks: true

install.tty

Definition: TTY device used for the console.

Example:

  1. install:
  2. tty: ttyS0,115200n8

install.vip

Definition: VIP of the Harvester management endpoint.

After installation, you can access the Harvester UI at https://<VIP>.

install.vip_mode

Definition: Mode of assigning the VIP.

Supported values:

  • dhcp: Harvester sends DHCP requests to get the VIP. You must specify the hardware address using the install.vip_hw_addr field.
  • static: Harvester uses a static VIP.

Example:

  1. install:
  2. vip: 192.168.0.100
  3. vip_mode: static

install.vip_hw_addr

Definition: Hardware address corresponding to the VIP.

You must configure an on-premises DHCP server to offer the configured VIP. This field is required when the value of install.vip_mode is dhcp. For more information, see Management Address.

Example:

  1. install:
  2. vip: 10.10.0.19
  3. vip_mode: dhcp
  4. vip_hw_addr: 52:54:00:ec:0e:0b

install.webhooks

Definition: Webhooks that allow you to receive notifications for certain installer-related events.

The installer sends HTTP requests to the specified URL. Multiple requests can be sent for a single event but if one request fails, the remaining requests are not sent.

Fields:

  • event: Event type that triggers an HTTP action on the webhook.
    • STARTED: The installation has started.
    • SUCCEEDED: The installation was completed without errors.
    • FAILED: The installation was unsuccessful.
  • method: HTTP method
  • url: URL to which HTTP requests are sent
  • insecure: When set to true, Harvester does not verify the server’s certificate. The default value is false.
  • basicAuth: When set to true, Harvester uses the “Basic” HTTP authentication scheme.
  • headers: When set to true, custom headers are included in the HTTP requests. Headers such as Content-Length are automatically included.
  • payload*: When set to true, payload data is sent with the HTTP requests. You may need to set the correct Content-Type header in the headers field to ensure that the server accepts the request.

Example:

  1. install:
  2. webhooks:
  3. - event: SUCCEEDED
  4. method: GET
  5. url: http://10.100.0.100/cblr/svc/op/nopxe/system/{{.Hostname}}
  6. - event: STARTED
  7. method: GET
  8. url: https://10.100.0.100/started/{{.Hostname}}
  9. insecure: true
  10. basicAuth:
  11. user: admin
  12. password: p@assword
  13. - event: FAILED
  14. method: POST
  15. url: http://10.100.0.100/record
  16. headers:
  17. Content-Type:
  18. - 'application/json; charset=utf-8'
  19. payload: |
  20. {
  21. "host": "{{.Hostname}}",
  22. "device": "hd"
  23. }

install.wipedisks

Definition: Setting that clears all disk partitions on the host using the sgdisk command.

install.with-net-images

Definition: Setting that determines if images are pulled from the internet after installation.

The value of this field is typically derived from the kernel parameter harvester.install.with_net_images. When the value is true, Harvester does not preload images packaged in the installation medium, and instead pulls images from the internet when necessary.


system_settings

Definition

You can overwrite the default Harvester system settings by configuring system_settings. See the Settings page for additional information and the list of all the options.

Harvester Configuration - 图8note

Overwriting system settings only works when Harvester is installed in “create” mode. If you install Harvester in “join” mode, this setting is ignored. Installing in “join” mode will adopt the system settings from the existing Harvester system.

Example

The example below overwrites containerd-registry, http-proxy and ui-source settings. The values must be a string.

  1. system_settings:
  2. containerd-registry: '{"Mirrors": {"docker.io": {"Endpoints": ["https://myregistry.local:5000"]}}, "Configs": {"myregistry.local:5000": {"Auth": {"Username": "testuser", "Password": "testpassword"}, "TLS": {"InsecureSkipVerify": false}}}}'
  3. http-proxy: '{"httpProxy": "http://my.proxy", "httpsProxy": "https://my.proxy", "noProxy": "some.internal.svc"}'
  4. ui-source: auto