Looks like i was quite lucky. At the moment, i was looking at the server notifications and fail2ban started screaming.
Almost 30 different IP addresses were blocked for ssh attack. And the locations are all around the world.
It was a server exposed online via some subdomain. Some ports were open, including 22. Is this something to be expected always?
What do the guy expect?
Does it make sense to report this to DigitalOcean as several of those IPs belong to DO?
118.45.151.148
125.91.123.149
43.134.180.30
128.199.208.187
43.133.33.240
43.163.218.44
43.156.238.11
129.226.91.96
43.156.240.201
43.134.33.175
43.153.226.222
43.134.231.46
43.154.189.227
159.223.74.41
156.232.11.117
156.232.13.213
43.134.132.76
43.153.202.243
43.134.230.140
43.156.101.180
64.227.176.121
43.159.40.202
124.156.2.182
146.190.142.125
139.59.160.73
49.51.183.1
68.168.132.152
94.72.4.20
103.180.149.5
Is it a good idea to have SSH open to the internet?
Yes there’s nothing wrong with it. SSH autt is robust.
As long as you’re running fail2ban there’s no harm in it. Without exception you should disable root login, and ideally you should disable password login and just use keys.
Fail2ban does all of nothing to protect you. At best it keeps the noise in the logs down a bit.
Competent attackers tend to use a botnet, blocking and rate limiting does jack shit against 10,000 IPs.
When I ran fail2ban, I modified the action to ban a much larger subnet instead of just one IP. I also banned it for 24 hours. Now I run OPNsense with geo blocking and just ignore the logs. It’s just noise.
It’s fine, but it’s a good idea to disable password authentication and only permit public key auth. Using a non-standard port helps reduce the spam in the logs a bit.