If you had told me 30 years ago that PHP would be responsible for billions of dollars of losses, I would have said, "I TOLD YOU SO!" Haven't run it since I was forced to write it, the worst thing I've ever written in.
Random high port will not make you any more secure.
You can use a tool like shodan or censys to quickly get a list of IP:ports with PHP <8.0.
Wireguard is UDP only. A UDP service doesn't have to answer invalid requests and Wireguard doesn't, unless you present the right key. This means it's basically impossible to tell an open port from one where no service is listening.
Depends on your firewall though. If you have a "default deny" then the random wireguard port will not send a RST package while the closed ports will. But UDP only sends a few RSTs per second so if you do a UDP-portscan, most systems will only send RSTs for the first few ports and then stop. If you scan very slowly you can find a wireguard port, but you can still not know it is wireguard, just that the port is actually open.
Reject is good for internal rules, but externally you should use drop all the time
Why a random high port? Wireguard doesn't respond on its port unless it gets its "handshake", so acts as stealth.
Because im trying to be cool.
But serious. If ever a wireguard vulnerability is found, which I is not very probable and a terrifying prospect, the default port will be the main target.
I use 443 udp, less public networks block it.
also you can be very edgy, e.g. 42069 or 57007 (0xDEAF)
I was a heavy stoner for years... I used 51420.
Funny is that once I forgot I configured that port and took me some hours of debugging to find out. :(
same here more or less, at home nothing is open anymore, everything goes through tailscale and a raspberry pi that also does the wake on lan for the big pc when I need it. the only thing exposed is a hetzner vps with a small browser game, and there I keep it boring, ufw with just 80 and 443 (and rtmp for a stream relay), caddy in front, the node app and postgres only reachable from inside the box. I guess that's the trade off that works for me, whatever strangers need to reach lives on one machine I watch closely, the rest stays behind the vpn
I do the same. It helps me sleep a little better
I do the same, SSH behind WireGuard and my public facing website/services are on a higher port number
The only traffic I see is from myself or the people I know personally. It prevents a hobby from turning into a job
However I do feel like the response to all these vulnerabilities are getting overblown. From my understanding, a lot of them require very specific conditions for exploits to work. It seems like AI pushers are trying to pad their security resumes with fancy websites. But I'm not a security expert so I could be wrong
I instead keep my system simple and mininal to avoid any headaches or at least keep them to a minimum
A lot of the CVEs lately have been rather overhyped, true, but malicious activity online has skyrocketed in the past month alone. I work in a sizeable SaaS shop and we're seeing a huge spike in curated highly intentful attacks on our systems.
The problem isn't with the CVEs themselves, it's more about response time now, bc any vulnerability, published or discovered, are getting exploited at faster rates than mitigation can keep up with. That window is the scary part.
It has made me paranoid enough to completely shut off port forwarding on my home router (I only had a handful being routed internally to Traefik). I'm now trying to find the time to properly set it up with cloudflared and mTLS... I'm tired guys 😭
I may be a bit sheltered. For the last 5 years I have been cutting myself off from subscription services and reduced my online/digital service needs. I won't say I predicted where we are today but I did see a trend not long after the COVID lockdowns and decided to reevaluate my expectations about technology and the internet. I've drastically lowered my expectations so I can still enjoy this technology a little bit even when everything seems so disappointing lately
I've also opted for extreme minimalism for my selfhosted server. Simple operating system, minimal programs/tools installed and no databases. I don't even use a git forge. Instead I clone my Codeberg repos to a basic file server run by my reverse proxy. I don't need any collaboration tools when all I want is my project repos to be accessible online. My future blog project will just be markdown files rendered to HTML to keep it as simple as can be (the inspiration). I've been so unstressed about all the new vulnerabilities that I've been peacefully enjoying the projections I am currently working on
I love the CVE's that go with physical access to the machine a ..... I mean if they have physical access to my machine they can do whatever they want anyways unless it is really locked down.
My server is a used Raspberry Pi 5 with an old USB3 portable harddrive. If anyone wanted root access, all they need is a Linux Mint live USB... So you're right, if they have physical access already, they can do whatever they like to it
The only really important information on there is the token for TLS certificate challenges. Everything else on there was hand selected to be shared knowledge. Who knows, maybe they can make use of all those survival books I have hosted from my basic little file server
Do you not encrypt your drives? Do you not use passwords? Do you not update your system/apps? I think this "if they have physical access it's already over" argument is overused. Depends on who is "they" and how is your computer/nas/server secured. Sure some protections are an overkill, but some basic measures can prevent a circumstantial crime to turn into something nastier (like impersonation or blackmail).
Doesn't matter. If an attacker has physical access, they can destroy everything.
You're still out.
For a CVE to state physical access is required means it's not significant, full stop.
As someone who's managed many servers/services, I've never in my career had physical access to the boxes I've used. I once was given a tour of one of our data centers, but only after special permission and a series of checks, and I was escorted the entire time.
Without physical security, you don't have any security
If an attacker has physical access, they can destroy everything.
That doesn't mean we should underestimate these CVEs, or the fixes. Blindly repeating what you say can do more harm than help. Maybe you know what you are doing (even then I don't agree with you) but your statements can do harm to people that don't (entirely). They skip on security patches, because they heard that its pointless and maybe 99.9 percent people will be okay, but how many won't?
Without physical security, you don’t have any security
As I said, not every "attacker" is intentional skilled hacker. And what if physical security is out of your hands. If you have server somewhere in "cloud" you have to trust someone is protecting the stuff. So what, you won't patch vulnerability because you are "sure" they have good physical security, or maybe then you take it seriously? What about social engineering? You can have good security measures but the people are always the weakest link. You may never be sure you are 100 % secure.
I agree. The fear helps their IPOs. We've always had bugs. I do the same and follow a KISS approach and do defense in depth.
I'm on the fence. I have auto updates on, I have my services routed through Cloudflare's proxy (besides Jellyfin) with rules blocking any connection from outside of my country, all running as rootless containers, all passing every security scan I've thrown at them... But I still have a sinking feeling something's gonna bite me.
But I really like being able to share links or connect without faffing with a VPN...
If you're using cloudflare tunnels, you might consider using mTLS, which effectively blocks (or provides a challenge, depending on how you configure your WAF rules) anything without your mTLS. The catch is that generally, apps need to support mTLS themselves, although this is not the case for browser based access.
I get that. I have my Nextcloud locked down pretty good. It's my only public facing service. But even with an A+ on Nextcloud's security scan and best practices on everything else, it still makes me a bit uncomfortable. Anything else goes over Wireguard. Everyone has a different level of comfort I guess. I'm a fairly careful individual when it comes to unrelated non-tech life decisions too. Some things just don't belong on the web... or in an AI infested operating system.
Every solution has its pros and cons. If you rely on something that requires public access (like randomly sharing access with others) than this is not a solution for you. With that said, you can go hybrid, anything high risk that you will never share with others (or only with a few predefined set of people that you can share a VPN config with) can go behind a VPN, anything you intend to share with others stays public but goes into a VM on a separate VLAN with restricted access to stuff to minimize the blast radius should sh*t go down. Personally this is what I'm trying to do, most things have been moved behind VPNs, only a few things remain public and I'm slowly moving them to separate VMs in a restricted network.
What made me change:
I can know that there is a php bug. But I still need for nextcloud to patch/update.
Same here, for quite a while. There is no reason for my NextCloud server to be exposed on the public internet. I still have access to it from everything via Tailscale and it works perfectly.
I already do this but I get nervus still with this second php thingy. Knock on wood that wireguard vulnerabities are not going to come into the thing.
Wireguard was actively designed to have fewer vulns. There's very little configuration available which dramatically reduces the number of scenarios a developer needs to consider when writing software. There are surely issues because no piece of software is perfect, but that was at least a stated design goal.
Wireguard is a relatively speaking new technology compared to PHP. Less legacy code to worry about.
I do not doubt there will/are vulnerabilities for it, but with any luck they will be few and far between with quick patches
Everything everywhere all at once, AI will be there to take away everything.
Acronyms, initialisms, abbreviations, contractions, and other phrases which expand to something larger, that I've seen in this thread:
| Fewer Letters | More Letters |
|---|---|
| Git | Popular version control system, primarily for code |
| SSH | Secure Shell for remote terminal access |
| TLS | Transport Layer Security, supersedes SSL |
| UDP | User Datagram Protocol, for real-time communications |
| VPN | Virtual Private Network |
| nginx | Popular HTTP server |
6 acronyms in this thread; the most compressed thread commented on today has 7 acronyms.
[Thread #173 for this comm, first seen 20th Nov 2026, 11:40] [FAQ] [Full list] [Contact] [Source code]
would you keep your doors unlocked and open 24/7? no? then why would you expose your server publicly?
to provide services???
This doesn’t make any sense lol
I only keep my server in a air tight safe, nothing in or out. It's the only way.
The keycloak/java one is a big deal since it was used to get into a few places.
Thanks god I’m not using any public-facing PHP applications anymore.
I fear this will eventually become the only way to safely selfhost for pretty much everyone. :/
It has always been the safest way to host private apps.
I don't mean WG specifically (although that's a great method) but I think everybody should be using some second layer that's enforced on a different level from the app itself: mTLS, OAuth, basic auth in the reverse proxy, even some geoblocking, anything.
With the recent nginx vulns I was just thinking about all the docker containers that are sitting there with a test webpage on an ancient version of nginx. I went and nuked all my test containers to reduce attack surface. But also sticking with VPN or Tailscale.
I’ve been doing this since I set up remote access. Nobody else should ever be using this but me, so why bother exposing anything?
Writing iptables isn’t even that hard
I'm honestly considering this for some of my small business customers too, expose their services other than their standard websites on high ports, that way it's generally only expected business contacts who will be making connections to those services.
AI has so many customers flapping about security at the moment.
That's why I opted for Headscale with a Tailscale fallback option. I'm thinking of setting up an old Pi to regularly check an HTML page on Neocities. If the page reads Heads, enable Headscale. If it reads Tails, switch to Tailscale. That way if my Headscale goes down while I'm away, I can change the page, what a bit, then SSH to the Pi via Tailscale and bounce the Headscale server. I can even power cycle the Headscale box via the LAN (Pi) thanks to some nifty JetKVM accessories.
Selfhosted
A place to share alternatives to popular online services that can be self-hosted without giving up privacy or locking you into a service you don't control.
Rules:
-
Be civil.
-
No spam.
-
Posts are to be related to self-hosting.
-
Don't duplicate the full text of your blog or readme if you're providing a link.
-
Submission headline should match the article title.
-
No trolling.
-
Promotion posts require active participation, with an account that is at least 30 days old. F/LOSS without a paywall has exceptions, with requirements. See the rules link for details. Tags [CBH] or [AIP] are required, see the links in Rule 8 for details.
-
AI-related discussions and AI-involved promotional posts have additional requirements for tagging, as noted in Rule 7 and the AI & Promotional Post Expanded Rules post, and find example disclosures here.
Resources:
- selfh.st Newsletter and index of selfhosted software and apps
- awesome-selfhosted software
- awesome-sysadmin resources
- Self-Hosted Podcast from Jupiter Broadcasting
Any issues on the community? Report it using the report flag.
Questions? DM the mods!