harvester-installer

Repo for building the Harvester ISO image. This includes the various scripts necessary to build the ISO itself, plus the harvester-installer binary and related scripts that perform system installation when the ISO is booted.

Building

To build an ISO image, run:

make

This will:

1. Build the harvester-installer binary.
2. Create an archive of all the necessary Harvester and Rancher charts and container images.
3. Create the harvester-cluster-repo container image, which provides a helm repository including the charts from the previous step.
4. Package everything from the above steps into an ISO image. The ISO image is built using Elemental Toolkit, and is based on harvester/os2, which in turn is based on SLE Micro.

The built ISO image is written to the dist/artifacts directory.

During the build, the harvester source and addons will be pulled automatically from https://github.com/harvester/harvester.git and https://github.com/harvester/addons.git resectively. If you would like to use an exiting local copy of either or both repositories instead, for example to pick up some development work in progress, you can do so as follows:

 $ export LOCAL_HARVESTER_SRC=/path/to/local/harvester/repo
 $ export LOCAL_ADDONS_SRC=/path/to/local/addons/repo
 $ make

Harvester Installation Process

Harvester can be installed by either booting the Harvester ISO, or via PXE Boot. When booting via ISO, harvester-installer runs interactively on the system console to allow you to configure the system. When booting via PXE, you don't get the interactive installer - instead you need to provide YAML files specifying the configuration to apply.

In both cases (ISO boot and PXE boot), the harvester-installer binary still runs in order to provision the system. This is put in place by system/oem/91_installer.yaml which in turn calls setup-installer.sh to start the installer on tty1.

When booted via ISO, the installer will prompt for configuration information (create a new cluster / join an existing cluster, what disks to use, network config, etc.). When booted via PXE, the kernel command line parameter harvester.install.automatic=true causes the interactive part to be skipped, and config will be retrieved from the URL specified by harvester.install.config_url.

The installer will run some preflight checks to ensure the system meets minimum hardware requirements. If any of these checks fail when run interactively, the first page of the installer will indicate which checks failed, and give you the option to proceed or not. When installing via PXE, if any checks fail, installation will abort and the failed checks will be visible on the system console, and also logged to /var/log/console.log in the installation environment. If you wish to bypass the preflight checks for testing purposes during automated installation, set the harvester.install.skipchecks=true kernel command line parmaeter.

Either way (ISO or PXE), the installer writes the final config out to a temporary file which is passed to harv-install which in turn calls elemental install to provision the system. The harv-install script also preloads all the container images. Finally, the system is rebooted.

On the newly installed system, harvester-installer remains active on the console in order to show the cluster management URL along with the current node's hostname and IP address.

Hacking the Interactive Part of harvester-installer

Ordinarily harvester-installer needs to be run from a booted ISO, so it can actually install a system. But, if you're only working on changes to the interactive part of the installer (e.g. adding or changing fields, or altering the workflow) and don't actually need to perform final installation, the binary can be quickly tested using vagrant with vagrant-libvirt. Here's how:

 $ USE_LOCAL_IMAGES=true make build
 $ vagrant up
 $ vagrant ssh
 > sudo DEBUG=true TTY=/dev/tty /vagrant/harvester-installer

Be sure the terminal window you use is fairly tall, or it will bail out with "panic: invalid dimensions" after you get past the networking screen. To break out of the installer, hit CTRL-C. If you rebuild the binary it can be sync'd back to a running vagrant box with vagrant rsync.

To remote debug the harvester-installer in your IDE, you need to install the delve package in your Vagrant box first.

 $ zypper addrepo https://download.opensuse.org/repositories/devel:/languages:/go/SLE_15_SP4/devel:languages:go.repo
 $ zypper refresh
 $ zypper install delve

Then rebuild the harvester-installer app and start it in the Vagrant box.

 $ REMOTE_DEBUG=true USE_LOCAL_IMAGES=true make build
 $ vagrant rsync
 $ vagrant ssh
 > sudo DEBUG=true TTY=/dev/tty dlv --listen=:2345 --headless=true --api-version=2 --accept-multiclient exec /vagrant/harvester-installer

The harvester-installer app will now listen for remote connections on port 2345.

Run vagrant ssh-config on your host to get the IP address that is needed to connect the IDE debugger to the harvester-installer app.

If you are using the Goland IDE, check out their documentation for information on how to attach to the harvester-installer process on the remote Vagrant box.

If you are using VS Code, create a remote debug configuration in .vscode/launch.json then press F5 to start debugging:

{
	"name": "Remote",
	"type": "go",
	"request": "attach",
	"mode": "remote",
	"port": 2345,
	"host": "IP_ADDRESS_OF_VAGRANT_BOX",
}

License

Copyright (c) 2024 Rancher Labs, Inc.

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at

http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.