If you’re self hosting Headscale you can configure your network such that Headscale is reachable on your network with or without internet access and available from the internet.

Don’t expect Gitea to make progress on federation. Forgejo is a fork of Gitea and anybody that cares about federation is probably on the Forgejo side of the fork.

If you’re running Kubernetes, what is the point of LXC or Proxmox in this setup? Kubernetes will give better scaling and utilization.

Git does have a server component. When git connects to an ssh remote it executes an ssh command that needs to be present.

You’re missing GitLab. I’d be looking at GitLab or Forgejo.

But you might not need this. When you access a private Git repository, you’re normally connecting over SSH and authenticating using SSH keys. By default, if you have Git installed on a server you can SSH to and you have a Git repository on that server in a location you can access, you can use that server as a Git remote. You only really want one these services if you want the CI pipelines or collaboration tools.

Port scanning isn’t abuse but automatically filing frivilous abuse reports is.

It’s not normal for - model-cache:/cache to be deleted on restart or even upgrade. You shouldn’t need to do this.

The server responds with a 404 error. If you’re using a reverse proxy, make sure the reverse proxy rules are right. Does it work when you connect directly?

It’s relatively easy for Cloudflare to profile clients as being web scrapers. A concerning amount of internet traffic goes through their servers in plain text.

Is this a lost karma bot?

A less intrusive solution would be to just put your sensitive data in LUKS and configure services that use that data to depend on the partition being mounted. That doesn’t require modifying the normal system startup process. You’re less likely to mess up your startup process at the expense of needing to be more mindful about where you’re putting your files.

Tang and Clevis have already been mentioned as a way for one server to boot using another server.

You can also create an environment where the server boots into a phase 1 where it obtains network connectivity and then waits for you to provide it the key to continue booting. The first phase is unencrypted, so don’t put sensitive data in there.

It is bad practice because of point 2 and if you have multiple replicas you can probably get different versions running simultaneously (never tried it). Get Rennovate. It creates PRs to increment the version number and it tries to give you the release notes right in the PR.

Borg / k8s / Docker are not the same thing. Borg is the predecessor of k8s, a serious tool for running production software. Docker is the predecessor of Podman. They all use containers, but Borg / k8s manage complete software deployments (usually featuring processes running in containers) while Docker / Podman only run containers. Docker / Podman are better for development or small temporary deployments. Docker is a company that has moved features from their free software into paid software. Podman is run by RedHat.

There are a lot of publicly available container images out there, and most of them are poorly constructed, obsolete, unreprodicible, unverifiable, vulnerable software, uploaded by some random stranger who at one point wanted to host something.

I did this but buying these on Amazon is scary. Try to find one that won’t burn your house down.

This fixes the issue of hosting any nefarious content.

How does removing images change anything? Any file can be transmitted by text, as we used to do with e-mail, and you don’t need to use images to make illegal or just intentionally offensive content.

Having a non-garbage domain provider can be a luxury. I used to work at a place where we were paying boatloads of money for certificates from Sectigo for internal services, and they were charging us extra per additional name and even more if we wanted a wildcard, even though it didn’t cost them anything to include those options. Getting IT to set up the DNS records for Let’s Encrypt DNS verification was never going to happen.

A large percentage of those hosts with SSH enabled are cloud machines because it’s standard for cloud machines to be only accessible by SSH by default. I’ve never seen a serious security guide that says to set up a VPN and move SSH behind the VPN, although some cloud instances are inherently like this because they’re on a virtual private network managed by the hosting provider for other reasons.

SSH is much simpler and more universal than a VPN. You can often use SSH port forwarding to access services without configuring a VPN. Recommending everyone to set up a VPN for everything makes networking and remote access much more complicated for new users.

Shodan reports that 35,780,216 hosts have SSH exposed to the internet.

Moving SSH to ports other than 22 is not security. The bots trying port 22 on random addresses with random passwords don’t have a chance of getting in unless you’re using password authentication with weak passwords or your SSH is very old.

SSH security updates are very infrequent and it takes practically no effort to keep SSH up to date. If you’re using a stable distribution, just enable automatic security updates.

Having SSH open to the internet is normal. Don’t use password authentication with weak passwords.