Harvester
Purpose: Rancher Harvester is an awesome tool that acts like a self-hosted cloud VDI provider, similar to AWS, Linode, and other online cloud compute platforms. In most scenarios, you will deploy "Rancher" in addition to Harvester to orchestrate the deployment, management, and rolling upgrades of a Kubernetes Cluster. You can also just run standalone Virtual Machines, similar to Hyper-V, RHEV, oVirt, Bhyve, XenServer, XCP-NG, and VMware ESXi.
:::note Prerequisites This document assumes your bare-metal host has at least 32GB of Memory, 200GB of Disk Space, and 8 processor cores. See Recommended System Requirements :::
First Harvester Node¶
Download Installer ISO¶
You will need to navigate to the Rancher Harvester GitHub to download the latest ISO release of Harvester, currently v1.1.2. Then image it onto a USB flashdrive using a tool like Rufus. Proceed to boot the bare-metal server from the USB drive to begin the Harvester installation process.
Begin Setup Process¶
You will be waiting a few minutes while the server boots from the USB drive, but you will eventually land on a page where it asks you to set up various values to use for networking and the cluster itself. The values seen below are examples and represent how my homelab is configured. - Management Interface(s): eno1,eno2,eno3,eno4 - Network Bond Mode: Active-Backup - IP Address: 192.168.3.254/24 <---- Note: Be sure to add CIDR Notation. - Gateway: 192.168.3.1 - DNS Server(s): 1.1.1.1,1.0.0.1,8.8.8.8,8.8.4.4 - Cluster VIP (Virtual IP): 192.168.3.251 <---- Note: See "VIRTUAL IP CONFIGURATION" note below. - Cluster Node Token: 19-USED-when-JOINING-more-NODES-to-EXISTING-cluster-55 - NTP Server(s): 0.suse.pool.ntp.org
:::caution Virtual IP Configuration The VIP assigned to the first node in the cluster will act as a proxy to the built-in load-balancing system. It is important that you do not create a second node with the same VIP (Could cause instability in existing cluster), or use an existing VIP as the Node IP address of a new Harvester Cluster Node. ::: :::tip Based on your preference, it would be good to assign the device a static DHCP reservation, or use numbers counting down from .254 (e.g. 192.168.3.254, 192.168.3.253, 192.168.3.252, etc...) :::
Wait for Installation to Complete¶
The installation process will take quite some time, but when it is finished, the Harvester Node will reboot and take you to a splash screen with the Harvester logo, with indicators as to what the VIP and Management Interface IPs are configured as, and whether or not the associated systems are operational and ready. Be patient until both statuses say READY. If after 15 minutes the status has still not changed to READY both for fields, see the note below. :::caution Issues with rancher-harvester-repo Image During my initial deployment efforts with Harvester v.1.1.2, I noticed that the Harvester Node never came online. That was because something bugged-out during installation and the rancher-harvester-repo image was not properly installed prior to node initialization. This will effectively soft-lock the node unless you reinstall the node from scratch, as the Docker Hub Registry that Harvester is looking for to finish the deployment does not exist anymore and depends on the local image bundled with the installer ISO.
If this happens, you unfortunately need to start over and reinstall Harvester and hope that it works the second time around. No other workarounds are currently known at this time on version 1.1.2. :::
Additional Harvester Nodes¶
If you work in a production environment, you will want more than one Harvester node to allow live-migrations, high-availability, and better load-balancing in the Harvester Cluster. The section below will outline the steps necessary to create additional Harvester nodes, join them to the existing Harvester cluster, and validate that they are functioning without issues.
Installation Process¶
Not Documented Yet
Joining Node to Existing Cluster¶
Not Documented Yet
Installing Rancher¶
If you plan on using Harvester for more than just running Virtual Machines (e.g. Containers), you will want to deploy Rancher inside of the Harvester Cluster in order or orchestrate the deployment, management, and rolling upgrades of various forms of Kubernetes Clusters (RKE2 Suggested). The steps below will go over the process of deploying a High-Availability Rancher environment to "adopt" Harvester as a VDI/compute platform for deploying the Kubernetes Cluster.
Provision ControlPlane Node(s) VMs on Harvester¶
Not Documented Yet
Adopt Harvester as Cluster Target¶
Not Documented Yet
Deploy Production Kubernetes Cluster to Harvester¶
Not Documented Yet