On the wire
hydra
Guesses a service's login by trying a list of usernames and passwords against it until one works.
Where nmap finds that a login service is open, hydra tries to get through it. Point it at a host and a
service (SSH, FTP, HTTP forms and many more), give it a username and a password list, and it tries each in turn,
reporting the pair that logs in. A weak password is one already on the list, so it falls in seconds. Nothing is broken
in the service: every guess is a real login attempt, so the run is loud — the service writes each one to its log.
Only guess logins you are allowed to; in netlab, the lab's own.
hydra command usage
hydra [-l USER | -L USERLIST] [-p PASS | -P PASSLIST] [-t TASKS] [-f] [-V] [-s PORT] TARGET SERVICE
Options
-
-l admin - one username to try (-L users.lst for a list of usernames)
-
-P pass.lst - a password list, one per line (-p for a single password)
-
-f - stop at the first working login, not after the whole list
-
-t 4 - run this many guesses at once (fewer is gentler, and steadier against SSH)
-
-V - print every username/password pair as it is tried
-
-s 2222 - the service listens on this port, not its default
hydra examples
-
hydra -l admin -P pass.lst 10.0.1.10 ssh - Tries every password in pass.lst against the admin account over SSH and names the one that works.
-
hydra -l admin -P pass.lst -t 4 -f 10.0.1.10 ssh - Runs four guesses at a time and stops at the first that logs in.
-
hydra -L users.lst -P pass.lst 10.0.1.10 ssh - Tries every username in one list against every password in another.
-
hydra -l admin -p password123 10.0.1.10 ssh - Tries a single password, to confirm one guess by hand.
-
hydra -l admin -P pass.lst ftp://10.0.1.10 - Guesses an FTP login, naming the service in the URL form instead.
Reading the output
$ hydra -l admin -P pass.lst -f 10.0.1.10 ssh [DATA] attacking ssh://10.0.1.10:22/ [22][ssh] host: 10.0.1.10 login: admin password: password123 1 of 1 target successfully completed, 1 valid password found
- The
[22][ssh]line names the login and password that worked: the weak password was on the list. 1 valid password found: the account is cracked. A long random password would have ended with0 valid passwords found.
Pitfalls
- Every guess is a real login the service logs: a dictionary run shows up as a burst of failed logins from one address, easy to alert on.