the best way to learn is by doing!

I just built my own automation around their official documentation; it’s fantastic.

https://www.wireguard.com/#conceptual-overview

https://www.wireguard.com/install/

https://www.wireguard.com/quickstart/

vyatta and vyatta-based (edgerouter, etc) I would say are good enough for the average consumer. If we’re deep enough in the weeds to be arguing the pros and cons of wireguard raw vs talescale; I think we’re certainly passed accepting a budget consumer router as acceptably meeting these and other needs.

Also you don’t need port forwarding and ddns for internal routing. My phone and laptop both have automation in place for switching wireguard profiles based on network SSID. At home, all traffic is routed locally; outside of my network everything goes through ddns/port forwarding.

If you’re really paranoid about it, you could always skip the port-forward route, and set up a wireguard-based mesh yourself using an external vps as a relay. That way you don’t have to open anything directly, and internal traffic still routes when you don’t have an internet connection at home. It’s basically what talescale is, except in this case you control the keys and have better insight into who is using them, and you reverse the authentication paradigm from external to internal.

Talescale proper gives you an external dependency (and a lot of security risk), but the underlying technology (wireguard) does not have the same limitation. You should just deploy wireguard yourself; it’s not as scary as it sounds.

Fail2ban and containers can be tricky, because under the hood, you’ll often have container policies automatically inserting themselves above host policies in iptables. The docker documentation has a good write-up on how to solve it for their implementation

https://docs.docker.com/engine/network/packet-filtering-firewalls/

For your usecase specifically: If you’re using VMs only, you could run it within any VM that is exposing traffic, but for containers you’ll have to run fail2ban on the host itself. I’m not sure how LXC handles this, but I assume it’s probably similar to docker.

The simplest solution would be to just put something between your hypervisor and the Internet physically (a raspberry-pi-based firewall, etc)

Devil’s Advocate:

How do we know that our brains don’t work the same way?

Why would it matter that we learn differently than a program learns?

Suppose someone has a photographic memory, should it be illegal for them to consume copyrighted works?

It’s fuckin’ art though

Oracle, SAP, Redhat, all of their customer portals require it for SSO. I’m not saying it should be that way, but it is.

I think you go about it the other way: break data analytics and advertising off from everything else. If every unit has to be self-sufficient without reliance on data collection and first-party advertising I think you fix most of the major issues.

I’m actually working on a vector DB RAG system for my own documentation. Even in its rudimentary stages, it’s been very helpful for finding functions in my own code that I don’t remember exactly what project I implemented it in, but have a vague idea what it did.

E.g

Have I ever written a bash function that orders non-symver GitHub branches?

Yes! In your ‘webwork automation’ project, starting on line 234, you wrote a function that sorts Git branches based on WebWork’s versioning conventions.

If there are any water pipes through the second half of the house you cannot let those exterior walls reach freezing temperatures. Whatever solution you go with needs to account for the entire space in some capacity.

+1 for cmk. Been using it at work for an entire data center + thousands of endpoints and I also use it for my 3 server homelab. It scales beautifully at any size.

That’s… not remotely true? Linux can absolutely install kernel drivers. If you mean running windows games under wine then sure, but then we’re no longer talking apples:apples. You could do the same thing on windows by running a game in a VM.

This is correct, as in windows a driver is the most straightforward method to runlevel0 access. It absolutely could at any time do exactly what crowdstrike did. But also so could Nvidia/amd with GPU drivers, your motherboard manufacturer with chipset and RGB drivers, etc. it’s not quite the smoking gun people make it out to be, as there are a lot of legitimate reasons to have this kind of system access.

The egregious part was that crowdstrike users agreed to allow a vendor to bypass canary channels and deploy straight to their endpoints.

Yes and no. In the best case, endpoints have enough cached data to get us through that process. In the worst case, that’s still a considerably smaller footprint to fix by hand before the rest of the infrastructure can fix itself.

With enough _autism in your overlay configs, sure, but in my environment tat leakage is still encrypted. It’s far simpler to just accept leakage and encrypt the OS partition with a key that’s never stored anywhere. If it gets lost, you rebuild the system from pxe. (Which is fine, because it only takes about 20 minutes and no data we care about exists there) If it’s working correctly, the OS partition is still encrypted and protects any inadvertent data leakage from offline attacks.

We do this in a lot of areas with fslogix where there is heavy persistent data, it just never felt necessary to do that for endpoints where the persistent data partition is not much more than user settings and caches of convenience. Anything that is important is never stored solely on the endpoints, but it is nice to be able to reboot those servers without affecting downstream endpoints. If we had everything locally dependant on fslogix, I’d have to schedule building-wide outages for patching.

Separate persistent data and operating system partitions, ensure that every local network has small pxe servers, vpned (wireguard, etc) to a cdn with your base OS deployment images, that validate images based on CA and checksum before delivering, and give every user the ability to pxe boot and redeploy the non-data partition.

Bitlocker keys for the OS partition are irrelevant because nothing of value is stored on the OS partition, and keys for the data partition can be stored and passed via AD after the redeploy. If someone somehow deploys an image that isn’t ours, it won’t have keys to the data partition because it won’t have a trust relationship with AD.

(This is actually what I do at work)

Endpoint is any PC/laptop/sign/POS/etc. It’s a catchall term for anything that isn’t a server. it basically refers to any machine that might be logged into and used by a non-IT user.

I think the vision was what Motorola delivered briefly a decade ago with webtop. The original version of it was a chrooted lubuntu with full access to apt, and custom applications that let you render your phone, or phone apps as an application. It was powerful enough to get me through my first 3 years of a computer science program in college with a lapdock as my primary “computer”. (Think a brainless laptop, that you dock your phone into)

https://arstechnica.com/gadgets/2011/03/motorola-atrix-the-ubuntu-powered-webtop-experience/

When they moved from android 2.3 to 4.0, they dropped the lubuntu webtop in favor of Android’s tablet mode, which was a huge bummer, and what made me get an actual laptop. Outside of gaming, if that were the average computer paradigm today I’d be a happy camper. Why buy two computers when you can buy one instead?