Introducing Hoster 🚀 - FreeBSD's VM and Jail management framework that will make your life easier. Whether you're an experienced sysadmin or just starting out, Hoster has got you covered.

Built using modern, rock solid and battle tested technologies like Go, FreeBSD, bhyve, ZFS, and PF, Hoster is a highly opinionated system that puts an emphasis on ease of use and speed of VM deployments. Whether you're managing a small home lab or a large-scale production, Hoster can easily accommodate your environment 🧑🏼‍💻

Here are some of the features you'll be able to use:

• PF firewall that works with every VM or Jail individually (you can use the VM and/or Jail names in the PF config directly - no need to explicitly implement VLANs to isolate VMs/Jails from one another)
• Native NAT management support using PF
• Bare-metal cloud-friendly deployment options (tested using Hetzner bare-metal cloud)
• Storage Dataset Encryption - your data is safe in the co-location, or on the bare-metal cloud
• Instant VM deployments - a new VM can be deployed in less than 1 second
• Cloud based VM images to avoid spending time with ISOs - it's very similar to docker pull, simply run hoster image download debian12 and it'll be ready for use in minutes (depending on your internet connection speed)
• CloudInit integration to help you forget about the manual VM configurations - simply deploy the VM, start it and it's immediately ready to be used (with the IP address configured, hostname set, and your own custom scripts executed upon first boot)
• Built-in and easy to use OpenZFS replication (based on OpenZFS's send/receive features), which gives you the ability to perform continuous asynchronous VM replication between the two or more hosts, to ensure the data safety and availability 🛡️
• RestAPI for ease of management, and to support the integration with 3rd party systems, or your own home-grown solutions
• An automated HA failover using the underlying RestAPI, so you can avoid the complex network configurations - it's all based on the HTTP protocol, which is easy to firewall and troubleshoot if there is a need for it
• PCI/GPU passthrough is supported, but considered experimental

To avoid any frustrations, here is the list of things NOT currently supported:

• Custom bhyve flags are not supported - I want to make sure every flag introduced has it's own config option
• File systems other than ZFS are not supported
• Linux as a host OS is not supported
• Very niche OSes are not supported due to some bhyve+Hoster limitations
• Only UEFI booting is officially supported, with just few exceptions for BIOS based Linux VMs
• Terraform is not supported - Hoster is too young to have any IaaC integrations at this point
• Custom binary files location and custom config files location is not supported - everything must reside within /opt/hoster-core to work properly (there is some WIP to overcome this limitation, but it's not ready yet)
• IPv6 is not supported yet (unless you want to manage it by hand, or sponsor us to speed up the IPv6 dev integration process)
• Nested virtualization is not supported by bhyve
• Code is not cross-platform - you can't run it on Illumos or any other BSD system, it only works on FreeBSD (and possibly some FreeBSD derivatives, like GhostBSD, HardenedBSD, etc, but it needs testing)
• bhyve doesn't support the live VM migration yet

Coming soon (or at least at some point 😄):

• Hyper Converged setup using HAST and OpenZFS (sync vs. async replication)
• Generally available WebUI for VM and Jail management
• The ability to specify a docker-compose.yaml file location during the VM deployment (hoster vm deploy --from-docker-compose compose.yaml) - it will be automatically picked up by docker and executed in a background tmux session (you can easily check if the docker deployment was successful using the tmux a command later on, when you SSH into the VM)
• Publicly available Grafana dashboards
• Prometheus integration - all VMs and Jails will be discovered and monitored automatically
• Automated, full mesh WireGuard config generation based on your internal cluster topology, with the support for multi-tenancy and multiple cluster groups
• Tailscale installation and configuration guide (specific to Hoster), which will provide an easy way to access and manage your cluster and it's resources
• LazyHoster (an interface similar to k9s or lazygit) - a CLI utility to control the whole fleet of machines over the REST API (for the WebUI haters 😎)
• More VM and Jail templates
• The ability to run Hoster in a Jail - this will allow to split a single hardware server into a multi-user isolated platform, with the ability to apply a highly granular control of just how much resources each user can utilize within their own, jailed, virtualization environment

Yes. Part of the project roadmap includes the development of a WebUI. The WebUI will serve as a user-friendly interface to interact with the system and control multiple hoster nodes simultaneously. While currently not the highest priority due to time constraints, I am open to exploring this feature further with increased community engagement and potential investment.

Our paying customers already have access to an early version of the WebUI, that looks like this:

The main idea behind our WebUI is to keep things simple. We are not aiming to be yet another XenSever/Proxmox feature clone: the WebUI will do basic things like managing and deploying new VMs, displaying monitoring information for the VMs and Hosts, managing VM snapshots, connecting to VNC, etc. Everything else in terms of configuration and Hoster management still happens on the CLI.

** Only production VMs will be included in the start-all, snapshot-all, replicate-all, etc

Whether you need a quick start guidance, or a deeper dive into Hoster's documentation, you can definitely do so by visiting this link: Hoster Core Docs