Harvester Overview

Harvester is an open-source hyper-converged infrastructure (HCI) software built on Kubernetes. It is an open alternative to using a proprietary HCI stack that incorporates the design and ethos of Cloud Native Computing.

harvester-ui

Harvester Features

Harvester implements HCI on bare metal servers. Harvester is designed to use local, direct attached storage instead of complex external SANs. It ships as an integrated bootable appliance image that can be deployed directly to servers through an ISO or PXE boot artifact.

Some notable features of Harvester include the following:

  1. VM lifecycle management including SSH-Key injection, cloud-init, and graphic and serial port console
  2. VM live migration support
  3. Supported VM backup, snapshot and restore
  4. Distributed block storage and storage tiering
  5. Multiple network interface controllers (NICs) in the VM connecting to the management network or VLANs
  6. Virtual Machine and cloud-init templates
  7. Rancher integration with multi-cluster management and the Harvester node driver
  8. PXE/iPXE boot support
  9. Virtual IP and bond NIC support
  10. Monitoring and logging integration

Harvester Architecture

The following diagram outlines a high-level architecture of Harvester:

Harvester Overview - 图2

  • Longhorn is a lightweight, reliable and easy-to-use distributed block storage system for Kubernetes.
  • KubeVirt is a virtual machine management add-on for Kubernetes.
  • Elemental for SLE-Micro 5.2 (based on openSUSE Leap 15.3 before v1.0.3) is an immutable Linux distribution designed to remove as much OS maintenance as possible in a Kubernetes cluster.

Hardware Requirements

To get the Harvester server up and running, the following minimum hardware is required:

TypeRequirements
CPUx86_64 only. Hardware-assisted virtualization is required. 8-core processor minimum for testing; 16-core or above preferred for production
Memory32 GB minimum; 64 GB or above preferred
Disk Capacity200 GB minimum for testing; 500 GB or above preferred for production
Disk Performance5,000+ random IOPS per disk (SSD/NVMe). Management nodes (first three nodes) must be fast enough for etcd
Network Card1 Gbps Ethernet minimum for testing; 10Gbps Ethernet recommended for production
Network SwitchTrunking of ports required for VLAN support

Quick Start

You can install Harvester via the ISO installation or the PXE boot installation. Instructions are provided in the sections below.

ISO Installation

You can use the ISO to install Harvester directly on the bare metal server to form a Harvester cluster. Users can add one or many compute nodes to join the existing cluster.

To get the Harvester ISO, download it from the Github releases.

During the installation, you can either choose to form a new cluster or join the node to an existing cluster.

  1. Mount the Harvester ISO disk and boot the server by selecting the Harvester Installer. iso-install.png

  2. Choose the installation mode by either creating a new Harvester cluster or by joining an existing one.

  3. Choose the installation device on which the Harvester cluster will be installed

    • Note: By default, Harvester uses GPT partitioning schema for both UEFI and BIOS. If you use the BIOS boot, then you will have the option to select MBR. iso-install-disk.png

  4. We recommend choosing a separate disk for storing VM data. iso-install-disk.png

  5. Configure the hostname and select the network interface for the management network. By default, Harvester will create a bonded NIC named mgmt-bo, and the IP address can be configured via DHCP or a statically assigned one (Note: The Node IP can not change at the lifecycle of a Harvester cluster. If DHCP is used, users must make sure the DHCP server always offers the same IP for the same Node. If Node IP is changed, the related Node cannot join the cluster and might even break the cluster).

    iso-installed.png

  6. Optional: Configure the DNS servers; use commas as delimiters.

  7. Configure the Virtual IP which you can use to access the cluster or join other nodes to the cluster (Note: If your IP address is configured via DHCP, you will need to configure static MAC-to-IP address mapping on your DHCP server in order to have a persistent Virtual IP, VIP must be different than any Node IP).

  8. Configure the cluster token. This token will be used for adding other nodes to the cluster.

  9. Configure the login password of the host. The default SSH user is rancher.

  10. We recommend configuring the NTP server to make sure all nodes’ times are synchronized. This defaults to 0.suse.pool.ntp.org.

  11. (Optional) If you need to use an HTTP proxy to access the outside world, enter the proxy URL address here. Otherwise, leave this blank.

  12. (Optional) You can choose to import SSH keys from a remote server URL. Your GitHub public keys can be used with https://github.com/<username>.keys.

  13. (Optional) If you need to customize the host with a Harvester configuration file, enter the HTTP URL here.

  14. Confirm the installation options and Harvester will be installed on your host. The installation may take a few minutes to complete.

  15. Once the installation is complete, the host will restart, and a console UI with management URL and status will be displayed. (You can use F12 to switch from the Harvester console to the Shell and type exit to go back to the Harvester console).

  16. The default URL of the web interface is https://your-virtual-ip. iso-installed.png

  17. Users will be prompted to set the password for the default admin user at first login. first-login.png

PXE/iPXE Installation

Harvester can also be installed automatically. Please refer to PXE Boot Install for detailed instructions and additional guidance.

Harvester Overview - 图9note

More iPXE usage examples are available at harvester/ipxe-examples.