I’ve been developing my own cloud platform that I’m calling “PiF” (Public Internet Fork) for the last few years with incus at the core. Many times I have come up against snags in development only to find that incus has shipped a new feature or fix matching my exact need at the same time. So, even though I’ve been mostly building in stealth - It feels like my efforts are in line with the needs of others. This makes me confident that my overall solution will be well received by the community.
That said, I still wanted to start sharing some of the things I’ve been working on in case anyone out there might be interested in becoming an early beta tester. The following utilize Incus’ powerful core functionality, incorporates many bleeding edge features, and elegantly wraps them together into one cohesive solution. I’m providing a broad description of a few things I have going so far:
S3 Native backups:
Automatically establishes a backup file on a remote S3 bucket for every launched container (via Restic). Stopping a container immediately backs it up to a remote S3 bucket. Starting a container first pulls the latest backup from S3. Stop a container on host A and start it on Host B. Your containers live on S3 and run on any host you want when needed.
Incus Billing System:
Show an hourly billing rate column in the “incus list” command. Define the monthly cost for your server, total available ram/cpu/ssd, and your profit margin - this gets reflected to an incus user as an hourly cost that changes based on the cpu/ram/ssd they provision to each instance. Every incus project gets its own “total hourly billing rate” that adds up all running instances together and a ledger file that defines their account balance/usages.
Incus multi-node mesh:
Automatically installs and configures headscale/tailscale (community version) onto every launched container. Ping containers by name and reach them regardless of them being local or on a remote node. “Incus list” shows containers from all hosts. Sidesteps incus’s native clustering functionality for a more scalable solution that still feels native
Production SSL, Auth, and reverse proxy:
Creates a wildcard SSL via lets encrypt, a Tinyauth server, and dedicated nginx instance for every incus project created. Every launched container automatically creates an nginx config for the container that points to container:8000. It also handles auth redirect as well. So when you launch a container named “containerA” and run an app on port 8000 - it’s automatically available at - https://containerA.project1.domain.com. Visiting the url first brings you to auth login and redirects you to the url after validation. Deleting a container deletes the associated nginx config.
If any of these features sound interesting, let me know and I can make it a priority to share demo videos and more info. If you have been working on similar things and need a hand, feel free to reach out. Thanks!
