So it’s my first time setting up a VPS. Is it to be expected to ban 54 IPs over a 12h timespan? The real question for me is whether this is normal or too much.
$ sudo fail2ban-client status sshd
Status for the jail: sshd
|- Filter
| |- Currently failed: 3
| |- Total failed: 586
| `- Journal matches: _SYSTEMD_UNIT=ssh.service + _COMM=sshd
`- Actions
|- Currently banned: 51
|- Total banned: 54
`- Banned IP list: [list of IPs]
fail2ban sshd.conf
$ sudo cat /etc/fail2ban/jail.d/sshd.conf
[sshd]
enabled = true
mode = aggressive
port = ssh
backend = systemd
maxretry = 3
findtime = 600
bantime = 86400
I have disabled SSH login via password. And only allow it over an SSH key.
$ sudo sshd -T | grep -E -i 'ChallengeResponseAuthentication|PasswordAuthentication|UsePAM|PermitRootLogin'
usepam no
permitrootlogin no
passwordauthentication no
You must log in or # to comment.
imagine the modern internet like a school yard of a very large school during recess, and it has just snowed.
Now imagine having ssh on port 22 as you a wearing a bright red wooly cap and standing in the middle of the yard jumping up and down and waving hands.
I personally wouldn’t waste your time with bans
Enable public key authentication and disable passwords
That’s very little actually
Move your SSH port from the standard 22 to one of the higher ones, like 53822
It’ll remove 99.something% of your attacks as nobody bothers with those ports.
Wouldn’t use a high port since they’re unprivileged.
so everyone can open them… so what? attacker who already gained local access can crash your original sshd and spin up his own one? admittedly a thinkable scenario… but can this even be abused in a pubkey auth scenario?
Mostly true, however the thing saving you would be host key verification, not pubkey authentication.
I’m just not into security by obscurity coupled with compromising the inbuilt mechanisms for making sure only root can open an SSHd.
Do you think high ports are irrelevant or only in this case for SSHd? If the former, why do you think the distinction exists in the first place?
I don’t see a reason to worry about that. Only matters if the machine is alreay compromised, and then it doesn’t matter either.
There are different levels of compromise: you could have local access or root access. This might allow a hacker to gain root access by faking an SSHd and asking for a password or something like that. Host key verification would save you in that case, but then again, there’s probably funny MITM things you can do with an existing SSHd.
Only 54?
Rookie numbers.
My favourite f2b rule is the one strike ban on SSH root login attempts. Any IP originating a SSH root login attempt is clearly compromised, and gets black holed on all my hosts for a month.
Note: direct SSH login isn’t permitted at all, the daemon is exposed purely to log the attempts.
a month.
I do it for a whole year. lol
The actual span is a random period between 2 and 4 weeks, it’s interesting to watch how long it takes for attempts to resume.
I prefer a more granular visibility, repeat offenders automagically ratchet up their stay in the sin bin.
They’re portscanning bots.
I made SSH IPv6-only and it stopped. You can’t scan IPv6 space for open ports.
You technically can scan IPv6
It is just less common
They can’t or they don’t?
My question too, pretty sure you can
The smallest possible subnet has 18.4 quintillion addresses.
You can’t scan it before encountering the heat death of the universe.
Outgoing connections are made on a different address that does not accept incoming connections. You never disclose your real IP when browsing.
So, no. It can’t be done.
That’s only if you use the brute force method
IPs are not secret and can be leaked in various ways. Don’t count on IPv6 as a security mechanism as that’s not what it was designed to do
It’s literally called “IPv6 privacy extension”. It’s what it does.
Unless you’re in the middle and fowarding the packets, you won’t stumble across a connectable IPv6 endpoint.
The IPv6 privacy extensions are something else entirely and are not used for anything server side
You should research IPv6 port scanning techniques. It is harder than IPv4 but still doable depending on the context. If your goal is simply to reduce network traffic due to bots that’s one thing. However, it should not be considered a security mechanism.
I have only one word to say : w00tw00t 😉
I deactivated the SSH daemon on my VPS and only use Tailscale SSH
Yeah. Sounds about right.
First of all disable root login over ssh. Second, move your ssh from port 22 to another port of your liking. Third, disable password authentication altogether and use only solid certificates.
Switching SSH to a non-standard port can cut down on log noise but it doesn’t really help with security. It’s trivial to identify ssh running on any port and attackers typically do full port scans anyway.
I’d put that effort towards allowlisting only trusted public ips or setting up wire guard/tailscale for ssh access instead.
Root login and password authentication are already disabled, and it’s very uncommon for self hosters to use SSH certificates at all.
Changing the SSH port away from 22 does not improve security unless your password is “password” or “admin”. Anybody who’s even slightly sophisticated will find your SSH service on the correct port and make requests there instead.
I don’t think changing the port has the intention of better security here, just reduces the amount of requests that dumb bots make. It’s noticable if you try it.
Yes, I meant ssh private/public keys, not certificates …
And changing the port reduces the amount of attempts in any case.
My SSH is only accessible thru a self hosted WireGuard VPN with a dedicated IP. Zero fails.
How do you deal with the risk of loosing the VPN (and thus the access)?
Yes, it is normal.
# fail2ban-client status sshd Status for the jail: sshd |- Filter | |- Currently failed: 10 | |- Total failed: 4433 | `- Journal matches: _SYSTEMD_UNIT=ssh.service + _COMM=sshd `- Actions |- Currently banned: 27 |- Total banned: 668 `- Banned IP list: 2.57.122.194 45.148.10.183 195.178.110.30 2.57.122.208 92.118.39.195 103.74.123.88 92.118.39.23 2.57.122.196 92.118.39.197 45.148.10.151 92.118.39.236 178.20.210.185 68.178.161.186 80.94.92.183 92.118.39.63 2.57.122.197 2.57.122.191 2.57.122.189 80.94.92.171 94.156.152.18 14.225.7.70 45.78.198.199 211.253.9.160 159.224.213.138 1.214.42.172 103.239.165.114 77.239.111.233That’s not a lot even. I’ve seen much worse. Also with password disabled. Some idiot bots still try to send it whole password lists.
Just move your SSH to another port, that is enough to get rid of most of the nuisance scans. Or allow SSH only on IPv6. That usually covers it (The IPv6 address space is too big to scan, unless you have a TLD directly pointing to your IPv6)
What I personally do is run an overlay VPN like tailscale and allow SSH in only via that.
What I personally do is run an overlay VPN like tailscale and allow SSH in only via that.
Same. I use Tailscale as an overlay on the pFsense box and the server itself.
Just 54? Those are rookie numbers bro. You need to open up a few more ports. LOL Honestly tho, seems pretty standard. You could change the SSH port which might lower some of the noise but bots now days are pretty sophisticated and it would be trivial to just scan your server and find out which port is SSH. If you want to stop tailing fail2ban nervously on the daily, you could use the hosts.allow/hosts.deny which would lock it down even further. Just remember to set host.allow first then host.deny. You could also deploy any number of secondary security packages like CrowdSec, Wazuh, et al.
I got 135 blocks via sshguard over the first 12 hours today. So, yeah, welcome to the Internet! 😄
