I understand that people enter the world of self hosting for various reasons. I am trying to dip my toes in this ocean to try and get away from privacy-offending centralised services such as Google, Cloudflare, AWS, etc.
As I spend more time here, I realise that it is practically impossible; especially for a newcomer, to setup any any usable self hosted web service without relying on these corporate behemoths.
I wanted to have my own little static website and alongside that run Immich, but I find that without Cloudflare, Google, and AWS, I run the risk of getting DDOSed or hacked. Also, since the physical server will be hosted at my home (to avoid AWS), there is a serious risk of infecting all devices at home as well (currently reading about VLANS to avoid this).
Am I correct in thinking that avoiding these corporations is impossible (and make peace with this situation), or are there ways to circumvent these giants and still have a good experience self hosting and using web services, even as a newcomer (all without draining my pockets too much)?
Edit: I was working on a lot of misconceptions and still have a lot of learn. Thank you all for your answers.
I have servers on Digital Ocean and Linode and also one in my basement, and have had no problem. I do have all services behind NPM (not to suggest it’s a panacea) and use HTTPS/SSH for everything. (not to suggest HTTPS/SSH are either) My use case could be different than yours - my immediate family are my only consumers - but have been running the same services in those locations for a few years now without issue.
Self hosting can save a lot of money compared to Google or aws. Also, self hosting doesn’t make you vulnerable to DDOS, you can be DDOSed even without a home server.
You don’t need VLANs to keep your network secure, but you should make sure than any self hosted service isn’t unnecessarily opens up tot he internet, and make sure that all your services are up to date.
What services are you planning to run? I could help suggest a threat model and security policy.
This is nonsense. A small static website is not going to be hacked or DDOSd. You can run it off a cheap ARM single board computer on your desk, no problem at all.
What?
I’ve popped up a web server and within a day had so many hits on the router (thousands per minute) that performance tanked.
Yea, no, any exposed service will get hammered. Frankly I’m surprised that machine I setup didn’t get hacked.
Don’t leave SSH on port 22 open as there are a lot of crawlers for that, otherwise I really can’t say I share your experience, and I have been self-hosting for years.
Am I missing something? Why would anyone leave SSH open outside the internal network?
All of my services have SSH disabled unless I need to do something, and then I only do it locally, and disable as soon as I’m done.
Note that I don’t have a VPS anywhere.
How do you reach into your server with SSH disabled without lugging a monitor and keyboard around?
My firewall, server, NAS and all my services have web GUIs. If I need SSH access all I have to do is enable it via web GUI, do what I need to, disable again.
If push comes to shove, I do have a portable monitor and a keyboard in storage if needed, but have not had the need to use them yet.
Some people want to be able to reach their server via SSH when they are not at home, but yes I agree in general that is not necessary when running a real home server.
Then use Wireguard to get into your local network. Simple as. All security risks that don’t need to be accessed by the public (document servers, ssh, internal tools, etc…) can be accessed via VPN while the port forwarded servers are behind a reverse proxy, TLS, and an authentication layer like Authelia/authentik for things that only a small group needs to access.
Sorry, but there is 1 case in 10000 where a home user would have to have publicly exposed SSH and 9999 cases of 10000 where it is not needed at all and would only be done out of laziness or lack of knowledge of options.
Yeah, I guess I’ve never needed to do that. That may change as I’m thinking of moving all my services from UnRaid to ProxMox to leave UnRaid for storage only.
I guess that’ll bring me back here soon enough.
Lol
You left stuff exposed is the only explanation. I’ve had services running for years without a problem
What class of IP was it?
I can’t say I’ve seen anything like that on the webservers I’ve exposed to the internet. But it could vary based on the IP you have if it’s a target for something already I suppose.
Frankly I’m surprised that machine I setup didn’t get hacked.
How could it if all you had was a basic webserver running?
I’ve been self-hosting a bunch of stuff for over a decade now, and have not had that issue.
Except for a matrix server with open registration for a community that others not in the community started to use.
Yes my biggest mistake was leaving a vps dns server wide open. It took months for it to get abused though.
If you are afraid of being ddosed which is very unlikely. Cloudflare has free ddos protection. You can put some but not all things behind their proxy.
Also instead of making things publicly available look in to using a VPN. Wireguard with “wireguard easy” makes this very simple.
VLANs do not make you network magically more secure. But when setup correctly can increase security a load if something has already penetrated the network. But also just to streamline a network and allow or deny some parts of the network.
Why would anyone ddos you? Ddos costs money andor effort. Noone is going to waste that on you. Maybe dos but not ddos. And the troll will go away after some time as well. There’s no gain in dosing you. Why would anyone hack your static website? For the lulz? If everything is https encrypted on your local net how does a hacker infest everything on your network?
DDOS can happen just from a script hammering on an exposed port trying to brute force credentials.
Then block them there are tools that restrict abuse
Acronyms, initialisms, abbreviations, contractions, and other phrases which expand to something larger, that I’ve seen in this thread:
Fewer Letters More Letters CGNAT Carrier-Grade NAT HTTP Hypertext Transfer Protocol, the Web HTTPS HTTP over SSL IP Internet Protocol NAS Network-Attached Storage NAT Network Address Translation NFS Network File System, a Unix-based file-sharing protocol known for performance and efficiency Plex Brand of media server package SSH Secure Shell for remote terminal access SSL Secure Sockets Layer, for transparent encryption TLS Transport Layer Security, supersedes SSL VPN Virtual Private Network VPS Virtual Private Server (opposed to shared hosting) nginx Popular HTTP server
13 acronyms in this thread; the most compressed thread commented on today has 8 acronyms.
[Thread #832 for this sub, first seen 26th Jun 2024, 10:25] [FAQ] [Full list] [Contact] [Source code]
Can you (or a human) expand NPM, presumably not the Node Package Manager?
NPM
Nginx Proxy Manager, I assume
It’s very possible. If you carefully manage your attack surface and update your software regularly, you can mitigate your security risks quite a bit.
The main problem is going to be email. I have found no reliable way to send email that does not start with “have someone else do it for you” or “obtain an IP block delegation”.
email isn’t that hard when you have a static IP, either from your network provider or via a VPS. Then, setup SPF, DKIM and DMARC and you’re good to go (at least for simple use cases like notifications. When you want to send out thousands of emails, you might need more.)
DDoS and hacking are like taxes: you should be so lucky as to have to worry about them, because that means you’re wildly successful. Worry about getting there first because that’s the hard part.
You don’t have to be successful to get hit by bots scanning for known vulnerabilities in common software (e.g. Wordpress), but OP won’t have to worry about that if they keep everything up to date. However, this is also necessary when renting a VPN from said centralised services.
Well he specified static website, which rules out WP, but yes. If your host accepts posts (in the generic sense, not necessarily specifically the http verb POST) that raises tons of other questions, that frankly were already well addressed when I made my post.
A static website and Immich
he specified static website, which rules out WP
Oops missed that
EDIT: And I missed Immich too
I feel like you have the wrong idea of what hacking acting a actually is… But yes, as long as you don’t do anything too stupid line forwarding all of your ports or going without any sort of firewall, the chances of you getting hacked are very low…
As for DDOSing, you can get DDOSed with or without self hosting all the same, but I wouldn’t worry about it.
Exactly piss off a script kiddie and get ddosed weather your self hosting or not.
One aspect is how interesting you are as a target. What would a possible attacker gain by getting access to your services or hosts?
The danger to get hacked is there but you are not Microsoft, amazon or PayPal. Expect login attempts and port scans from actors who map out the internets. But I doubt someone would spend much effort to break into your hosts if you do not make it easy (like scripted automatic exploits and known passwords login attempts easy) .
DDOS protection isn’t something a tiny self hosted instance would need (at least in my experience).
Firewall your hosts, maybe use a reverse proxy and only expose the necessary services. Use secure passwords (different for each service), add fail2ban or the like if you’re paranoid. Maybe look into MFA. Use a DMZ (yes, VLANs could be involved here). Keep your software updated so that exploits don’t work. Have backups if something breaks or gets broken.
In my experience the biggest danger to my services is my laziness. It takes steady low level effort to keep the instances updated and running. (Yes there are automated update mechanisms - unattended upgrades i.e. -, but also downwards compatibility breaking changes in the software which will require manual interactions by me.)
All reverse proxies i have used do rudimentary DDoS protection: rate limiting. Enough to keep your local script kiddy at bay - but not advanced stuff.
You can protect your ssh instance with rate limiting too but you’ll likely do this in the firewall and not the proxy.
…maybe use a reverse proxy…
+1 post.
I would suggest definitely reverse proxy. Caddy should be trivial in this use case.
cheers,
Reverse proxies don’t add security.
deleted by creator
A reverse proxy is used to expose services that don’t run on exposed hosts. It does not add security but it keeps you from adding attack vectors.
They usually provide load balancing too, also not a security feature.
Edit: in other words what he’s saying is true and equal to “raid isn’t baclup”
I don’t get why they say that? Sure, maybe the attackers don’t know that I’m on Ubuntu 21.2 but if they come across https://paperless.myproxy.com and the Paperless-NGX website opens, I’m pretty sure they know they just visited a Paperless install and can try the exploits they know. Yes, the last part was a bit snarky, but I am truly curious how it can help? Since I’ve looked at proxies multiple times to use it for my selfhosted stuff but I never saw really practical examples of what to do and how to set it up to add an safety/security layer so I always fall back to my VPN and leave it at that.
Not every path is mapped with the reverse proxy.
I’m positive that F5’s marketing department knows more than me about security and has not ulterior motive in making you think you’re more secure.
Snark aside, they may do some sort of WAF in addition to being a proxy. Just “adding a proxy” does very little.
So, you’ve gone from:
reverse proxies don’t add security
to:
“adding a proxy” does very little
What’s next?
Give up. You don’t know what the fuck you’re talking about.
… You’re joking right?
No.
May not add security in and of itself, but it certainly adds the ability to have a little extra security. Put your reverse proxy in a DMZ, so that only it is directly facing the intergoogles. Use firewall to only expose certain ports and destinations exposed to your origins. Install a single wildcard cert and easily cover any subdomains you set up. There’s even nginx configuration files out there that will block URL’s based on regex pattern matches for suspicious strings. All of this (probably a lot more I’m missing) adds some level of layered security.
Put your reverse proxy in a DMZ, so that only it is directly facing the intergoogles
So what? I can still access your application through the rproxy. You’re not protecting the application by doing that.
Install a single wildcard cert and easily cover any subdomains you set up
This is a way to do it but not a necessary way to do it. The rproxy has not improved security here. It’s just convenient to have a single SSL endpoint.
There’s even nginx configuration files out there that will block URL’s based on regex pattern matches for suspicious strings. All of this (probably a lot more I’m missing) adds some level of layered security.
If you do that, sure. But that’s not the advice given in this forum is it? It’s “install an rproxy!” as though that alone has done anything useful.
For the most part people in this form seem to think that “direct access to my server” is unsafe but if you simply put a second hop in the chain that now you can sleep easily at night. And bonus points if that rproxy is a VPS or in a separate subnet!
The web browser doesn’t care if the application is behind one, two or three rproxies. If I can still get to your application and guess your password or exploit a known vulnerability in your application then it’s game over.
The web browser doesn’t care if the application is behind one, two or three rproxies. If I can still get to your application and guess your password or exploit a known vulnerability in your application then it’s game over.
Right!?
Your castle can have many walls of protection but if you leave the doors/ports open, people/traffic just passes through.
So I’ve always wondered this. How does a cloudflare tunnel offer protection from the same thing.
They may offer some sort of WAF (web application firewall) that inspects traffic for potentially malicious intent. Things like SQL injection. That’s more than just a proxy though.
Otherwise, they really don’t.
I have a dozen services running on a myriad of ports. My reverse proxy setup allows me to map hostnames to those services and expose only 80/443 to the web, plus the fact that an entity needs to know a hostname now instead of just an exposed port. IPS signatures can help identify abstract hostname scans and the proxy can be configured to permit only designated sources. Reverse proxies also commonly get used to allow for SSL offloading to permit clear text observation of traffic between the proxy and the backing host. Plenty of other use cases for them out there too, don’t think of it as some one trick off/on access gateway tool
My reverse proxy setup allows me to map hostnames to those services and expose only 80/443 to the web,
The mapping is helpful but not a security benefit. The latter can be done with a firewall.
Paraphrasing - there is a bunch of stuff you can also do with a reverse proxy
Yes. But that’s no longer just a reverse proxy. The reverse proxy isn’t itself a security tool.
I see a lot of vacuous security advice in this forum. “Install a firewall”, “install a reverse proxy”, etc. This is mostly useless advice. Yes, do those things but they do not add any protection to the service you are exposing.
A firewall only protects you from exposing services you didn’t want to expose (e.g. NFS or some other service running on the same system), and the rproxy just allows for host based routing. In both cases your service is still exposed to the internet. Directly or indirectly makes no significant difference.
What we should be advising people to do is “use a valid ssl certificate, ensure you don’t use any application default passwords, use very good passwords where you do use them, and keep your services and servers up-to-date”.
A firewall allowing port 443 in and an rproxy happily forwarding traffic to a vulnerable server is of no help.
They’re a part of the mix. Firewalls, Proxies, WAF (often built into a proxy), IPS, AV, and whatever intelligence systems one may like work together to do their tasks. Visibility of traffic is important as well as the management burden being low enough. I used to have to manually log into several boxes on a regular basis to update software, certs, and configs, now a majority of that is automated and I just get an email to schedule a restart if needed.
A reverse proxy can be a lot more than just host based routing though. Take something like a Bluecoat or F5 and look at the options on it. Now you might say it’s not a proxy then because it does X/Y/Z but at the heart of things creating that bridged intercept for the traffic is still the core functionality.
You can’t port map the same port to different services on a firewall. Reverse proxy lets you open one port and have multiple services on it. Firewall can protect exposed services one I geoip block every country but my own two use crowded to block what they consider malicious ips.
+1 for the main risk to my service reliability being me getting distracted by some other shiny thing and getting behind on maintenance.
I’m in this comment.
It’s crowded.
Hey no to be harsh or anything but did you actually made your research? Plenty of people self host websites on their house without AWS , Google or Cloudfare and it works fine.
Not my experience so far with my single service I’ve been running for a year. It’s making me even think of opening up even more stuff.
Yeah na, put your home services in Tailscale, and for your VPS services set up the firewall for HTTP, HTTPS and SSH only, no root login, use keys, and run fail2ban to make hacking your SSH expensive. You’re a much smaller target than you think - really it’s just bots knocking on your door and they don’t have a profit motive for a DDOS.
From your description, I’d have the website on a VPS, and Immich at home behind TailScale. Job’s a goodun.
Just changing the SSH port to non standard port would greatly reduce that risk. Disable root login and password login, use VLANs and containers whenever possible, update your services regularly and you will be mostly fine
Other than the low chance of you being targeted I would say only expose your services through something like Wireguard. Other than the port being open attackers won’t know what it’s for. Wireguard doesn’t respond if you don’t immediately authenticate.
It depends on what your level of confidence and paranoia is. Things on the Internet get scanned constantly, I actually get routine reports from one of them that I noticed in the logs and hit them up via an associated website. Just take it as an expected that someone out there is going to try and see if admin/password gets into some login screen if it’s facing the web.
For the most part, so long as you keep things updated and use reputable and maintained software for your system the larger risk is going to come from someone clicking a link in the wrong email than from someone haxxoring in from the public internet.
