To mitigate the effort to maintain my personal server, I am considering to only expose ssh port to the outside and use its socks proxy to reach other services. is Portknocking enough to reduce surface of attack to the minimum?
In your SSH config, you should disallow root login and password authentication.
It is more secure than these tommyknockers :-) but you can do that additionally, if you feel like it.
Why disallow root login? I always need root when I connect, and stealing the password by aliasing sudo/doas is trivial. It seems to me it would just make life harder for no benefit.
Because then:
- you also need to know the correct username
- audits and logging shows which user used sudo to gain root access
- you also need to know the correct username
Use a secure password or key. Security by obscurity is no security.
- audits and logging shows which user used sudo to gain root access
That is not the point that was made. Once access to sudo or root you already have lost.
You’re making it that much easier for someone to brute force logging in or to exploit a known vulnerability. If you have a separate root password (which you should) an attacker needs to get through two passwords to do anything privileged.
This has been considered an accepted best practice for 20+ years and there’s little reason not to do it anyways. You shouldn’t be running things as root directly regardless.
When you have secure passwords kr key auth. Brute force is not a problem. What vulnerability are you talking about? Complete auth bypass? Then the username would be no problem either since you can just brute force usernames.
Heartbleed was a thing that happened.
Security though obscurity, BY ITSELF, is not security. But it’s great at slowing attackers and thwarting automated scripts.
It’s bad security to ignore possible mitigations to a problem just because it isn’t as full fix.
Defense in depth.
Anyone who’s certified NET+ or higher knows this.
That video here: https://youtu.be/fKuqYQdqRIs?feature=shared
Does a great job actually explaining a lot of my points. And it is produced by an actual security auditor and researcher. Just because everyone is doing it does not make it a security benefit that matters.
Here is an alternative Piped link(s):
https://piped.video/fKuqYQdqRIs?feature=shared
Piped is a privacy-respecting open-source alternative frontend to YouTube.
I’m open-source, check me out at GitHub.
Why disallow root login?
It is very easy to throw a dictionary at your port 22. It happens every few minutes. And they all try it with the username=root unless they know something better.
they need to guess a username i assume
just use tailscale
I would only expose a VPN and use that to access the other services.
Why? There’s no downside to ssh, if anything it’s easier to set up.
A VPN would give you access to a network, but not necessarily the devices on that network. It adds another layer of security as the user not only has to have SSH credentials/keys, but they also have to have the same for the VPN. SSH and VPNs would really be used in conjunction with each other.
It’s onion security.
If you only want to provide ssh access to one host, sure. If you want to provide other services, on multiple hosts, then you’re either making it a jump box or a proxy, while a VPN would provide direct access (or at least as defined in the firewall and routing rules).
Removed by mod
Other then the slowly increasing log file (if you use fail2ban for example), it will take thousands of years to actually hack you through this method as long as root auth is disabled and authentication is only via SSH keys, I wouldn’t worry about it.
It is possible to tighten the security of a machine to the point it is no longer usable. It is important to secure our devices but we cannot forget about convenience, so the trick is to tighten it but also make it so you don’t have to jump through a number of hoops till you get to your destination.
I for example, wouldn’t use your method because it would make it difficult to use some services I host from my phone.
Port knockers for the most part aren’t worrying. In an ideal situation, the only ports that should be open are 22, 80, 443 and using a reverse proxy to mask headers. (Poor configuration for example, go to Shodan and type bitwarden in the search bar and see how many people expose their instances to the world carelessly without an SSL cert) and the occasional UDP for game servers/media servers.
I used to SSH into my server and proxy out from there. Then I learned how shit of a solution that is for daily use and set up a vpn like a normal person.
Acronyms, initialisms, abbreviations, contractions, and other phrases which expand to something larger, that I’ve seen in this thread:
Fewer Letters More Letters SSH Secure Shell for remote terminal access SSL Secure Sockets Layer, for transparent encryption UDP User Datagram Protocol, for real-time communications VPN Virtual Private Network VPS Virtual Private Server (opposed to shared hosting)
[Thread #128 for this sub, first seen 10th Sep 2023, 16:25] [FAQ] [Full list] [Contact] [Source code]
I use a vps running openvps. Then i only allow ssh access from my vpn ip address.
Sounds like security through obscurity to me.
Highly susceptible to replay and man in the middle attacks.
If you’re gonna combine that with another authentication method (and you should), then I see little advantage over just going with the other auth method.
Sure? It certainly detracts bots that now don’t discover the SSH port anymore. Against a targeted attack it’s less useful, but that is a very hard problem in any case. If someone is out to get you specifically, it will be a tough battle.
Bots do not matter. They try just common know exploits. If your root password is not root you are fine.
Root login should be disabled, and ideally remote user auth should be key only, not password. And you should have a passphrase on your key.
Should be
Why? Dont recite a blogpost to me explain it. Following blindly security practices you do not understqnd can be very dangerous.
Disableing the root login gains nothing in regarding security. If you have a secure key or a passwordthey attacker will not get in no matter what. And once a account is compromised it ia trivial to extract the sudo passwors with simple aliases.
Passwords can be as secure as keys. Yes be default a weak key is still more secure then a weak passwors. But if you have a strong password policy in place it does not matter. Most valid argument for keys is the ease of you
Having a passphrase on the key is for example for my usecase irrelevant. I run full disk encryption on every device. A passphrase on those keys would not gain me much security only more inconvenience.
Disableing the root login gains nothing in regarding security.
This is usually not the reason people recommend disabling root login. Since root is an anonymous account not tied to an actual person, in a corporate setting, you do not really know who used that account if you allow root login. If this is relevant for a personal home network is for you to decide. I would say there is not such a strong argument for it to be made in that setting.
I absolutely agree with your. It can makes sence the disable it for access control, loging, auditing, etc. .
But when you look online or just in the comment section here lots of ppl actually recommend it as a security meassure against attackers. “Need to brute force the username as well”
If you’re worried about bots just use a non-standard port and move on. I did that on my own VPS just to cut down on log chatter and I get absolutely zero ssh attack attempts after the change.
Highly susceptible to replay and man in the middle attacks.
fwknop
isn’t susceptible to either.
Instead of ssh I use wireguard directly. It’s a simple protocol based on public/private keys with great performance and security.
Wireguard is stateless and establishes connections really quickly on demand. This means the battery isn’t impacted even though it’s always on, since the VPN doesn’t have to maintain a constant connection. At least that’s the case if your routing only a specific subnet (e.g. 192.168.1.0/24 and not all traffic through it 0.0.0.0/0).
What kind of port knocking just going to ports in sequence? Or someone wrote one that looks for a key signed and is supposedly not replayable.