I accidentally attempted to SSH into one of my servers from a device that did not contain my ssh key. I configure all of my servers to only allow authentication via cryptographic keys. Root ssh as well as password auth are disabled.
To my surprise, I was able to log in to my server with a password despite this. Baffled, I first tried some other servers. 2 of the 5 other servers I tried were accessabke via password.
After some swift investigation the culprit was found, a cloud-init ssh config in sshd_config.d/ with one line: password_authentication Yes.
So TLDR PSA…if you run a server in any type of virtualized environment, including a VPS, check your /etc/ssh/sshd_config.d/ folder. And more broadly, actually thoroughly test your ssh access to confirm everything is working as you intend it to.
Show your effective sshd server config:
sudo sshd -TYeah that sounds pretty bad. Is there a quick way to disable ssh keys to test?
/etc/ssh/ssh.d/
You mean
/etc/ssh/sshd_config.d?Fixed it. Thanks
You for got the
d!sshd_config
picard_facepalm.png. can you tell I just Tab through terminal?
I could even go further into saying: always test every change you make, do not assume the change has been made because you updated a file.
Good advice. One should always test, for correctness, not just infer.
OpenSSH right? What version?
No issues with DropbearThis is good advice in general. Think of it like penetration testing. You really should verify what you can actually access remotely on a device and not assume you have any level of protection until you’ve tried it.
Log files can also contain signs of attack like password guessing. You should review these on a regular basis.
Had a similar issue with tlp recently. I just happened to notice the laptop battery was at 100%, and said it was charging. I double and triple checked the config file, but the tlp-stat -b still showed the thresholds at 90%-100%.
Turns out tlp, at some point, started ignoring /etc/tlp.conf, and was pointing to /etc/default/tlp
