scans for open ports ran continuously since the 1990s, it was never a big deal. Also they only run on lower ports (not that it matters)

what are you talking about killing your internet performance? You can have hundreds of thousands of scans per day (which isn’t gonna happen, you won’t even get 100) and it still won’t bog down jank cable internet from early 2000s

Do it.

There’s really not much that can end badly, someone gets in your network (unlikely anyone even knows it exists)? reformat all your shit. Just by knowing what a DMZ is you are already more qualified than half the people I’ve met self hosting

do you run a business out of your house? do you run a bunch of peoples personal info? does anyone else? If you answered no to all of these then there really isn’t much that can “go wrong” you can just unplug your shit.

hosting email also isn’t that big of a deal but your home ISP will block port 25, you need to have a “business” one for them to unblock it and even then sometimes have to directly request it. Things like mailcow docker make it dead easy.

and yea as the other guy said always update your stuff

I agree. Just run it. that’s how I learned decades ago. Don’t ignore it either if you wanna get better.

The risks are just as bad as owning some amazon IoT device

yea they’ll plug in the drive and windows will popup “this drive needs to be formatted” and the rest is history

Yea, would probably call for integrating the actual issues into git too lol

not sure if there’s a docker specific one, I mostly avoid docker. the volumes are accessible by anything as far as I’ve seen though so anything that does incremental backups (borg, restic, rsync, rsnapshot) should be able to do it there too?

Lol yea I figured it was some governmental reason but I see name.com etc has .be so no idea why they wouldn’t have it. Some TLDs have specific restrictions and only allow certain sites to resell it

they don’t have to figure it out, you are the one running it

If you’re doing incremental backups it should be nearly instant if nothing changes. But yea ignoring garbage like cache is always nice

I don’t consider it excessive since stuff like rsync -a --inplace --delete on a giant directory will be nearly instant if nothing changed, even with terabytes of data (of smallish files) since it isn’t going to copy anything except changed things.

borg used to have an issue where it took forever even with nothing changed but it looks like it is mostly fixed so not an issue now. I moved to restic backup years ago though (which is basically borg without those issues) also proxmox has the proxmox backup server that does similar for all the vps/containers

been running it 7 years and it really isn’t that much of an issue. Conduit unfortunately likely has a lot of things unseen since the user base is so small.

I still wish migration was easy because I’d change to dendrite

Yea you may lose performance or may not, depends how the chip works on the actual pcie card. (Even if the slot has more than enough bandwidth at x1)

You can easily do a performance test after plugging it in though. Typically even an x8 interface will work with just x1 pcie connectors just slower. Even if that x1 interface is several generations later and has more bandwidth than that x8 needs lol

some ramblings of why just having the physical slots doesn’t always mean ___ can be found on this completely unrelated repo: https://github.com/magic-blue-smoke/Dual-Edge-TPU-Adapter (might be in an issue thread too unfortunately but he has more info here on how bandwidth works at that level than any other sites)

call them up and ask them why they don’t lmao. I actually did have to call them once for a company thing and someone in Oregon picked up… Unheard of service.

Yea with that activity comes hosting cost and it has to come from somewhere because 99% of them wouldn’t run their own instance

porkbun (USA) generally has the best prices and will give you a massive list of everything as well which most other sites won’t. I generally get my sites for 10 years at a time ~ $2 per year (whenever some random TLD is on sale)

https://en.wikipedia.org/wiki/Gandi

This one is in the top registrars ever to exist and is very old as well. Coincidentally also not in USA. Other rivals of “greatness” are things like porkbun (USA) and namecheap (USA)

Looks like other users say gandi has gone downhill. Couldn’t recommend any over porkbun though so good luck.

What is your reason to blog, for yourself? If so I run (used to run gitea) forgejo and just spam everything in issue threads on specific repositories

this uses 150mb of ram, basically 0% cpu

forgejo is actively working on federation, it is there not sure how done it is I don’t use it.

restic also written to do what borg does but fix a bunch of the shortcomings that they also now fixed (many of) in the new 2.0/2.x versions

just run wireguard on the jelly server…

wireguard honestly takes like 30 seconds to do once you learn how to use it.

I wouldn’t say “great” it’s ok software. Not even due to all of those security things which is a nightmare too. They do things like break the search speed months ago and not have any idea why, it’s so insanely slow and on top of that it somehow lags the entire client when searching too, not just the server which is the only thing doing the query. Lots of issues just with that.