It buys you enough time to check the journals and see that a group of IPs have attempted various ports giving you enough time to block the IP altogether.
It also buys you disinterest from the malicious host, since probably there’s a hard limit on how many ports they will test, and they will flag your machine as “too much work” and try another.
Again, I agree with you that obfuscation is not security, but it sure does help.
From what I understand you obfuscate the port in order to limit the amount of incoming attacks. But then fail2ban would be a much more effective tool.
The disinterested aspect you described is the actual problem. Because it’s based on the assumption your port won’t be found, but it definitely will, and as soon as that happens you’ll end up in a database such as shodan and the entire effect is GONE.
Kind of - I obfuscate the port because it won’t show up on the list of well known TCP and UDP ports[0], but be in ephemeral region. To attack this port they would have to guess the number or do a full wide port scan of the system which will waste a large amount of their time. Though granted, they need probably less than a week.
I’ve honestly never understood the defaults of fail2ban, which seemed to do nothing on every system I’ve tried it on. I get much better results by parsing the journalctl logs, and grouping the ips and then passing them directly into iptables or UFW.
I hear you, but I disagree:
It buys you enough time to check the journals and see that a group of IPs have attempted various ports giving you enough time to block the IP altogether.
It also buys you disinterest from the malicious host, since probably there’s a hard limit on how many ports they will test, and they will flag your machine as “too much work” and try another.
Again, I agree with you that obfuscation is not security, but it sure does help.
From what I understand you obfuscate the port in order to limit the amount of incoming attacks. But then fail2ban would be a much more effective tool.
The disinterested aspect you described is the actual problem. Because it’s based on the assumption your port won’t be found, but it definitely will, and as soon as that happens you’ll end up in a database such as shodan and the entire effect is GONE.
Kind of - I obfuscate the port because it won’t show up on the list of well known TCP and UDP ports[0], but be in ephemeral region. To attack this port they would have to guess the number or do a full wide port scan of the system which will waste a large amount of their time. Though granted, they need probably less than a week.
I’ve honestly never understood the defaults of fail2ban, which seemed to do nothing on every system I’ve tried it on. I get much better results by parsing the journalctl logs, and grouping the ips and then passing them directly into iptables or UFW.
You’re probably right. What is shodan?
0: https://en.wikipedia.org/wiki/List_of_TCP_and_UDP_port_numbers