How should I defend against this type of attack? This is a tightvnc logfile excerpt from a linode cloud server running Ubuntu 12.04 (Precise) with the ubuntu-desktop package added to the bare server. By default this Ubuntu linode cloud server has exactly one user named root. No other users will ever be needed for this server and sudo is considered to be inconvenient. This server runs exactly one application for a very small anonymous business. When running, the application talks unattended through a socket/port over the internet. The viewing of application results occurs through an SSH tunnel for VNC. Light administration occurs via ssh, sftp, scp and rsync with SSH.
The IPs shown in the excerpt are not the IP of the legitimate client.
05/06/12 20:07:32 Got connection from client 69.194.204.90
05/06/12 20:07:32 Non-standard protocol version 3.4, using 3.3 instead
05/06/12 20:07:32 Too many authentication failures - client rejected
05/06/12 20:07:32 Client 69.194.204.90 gone
05/06/12 20:07:32 Statistics:
05/06/12 20:07:32 framebuffer updates 0, rectangles 0, bytes 0
05/06/12 20:24:56 Got connection from client 79.161.16.40
05/06/12 20:24:56 Non-standard protocol version 3.4, using 3.3 instead
05/06/12 20:24:56 Too many authentication failures - client rejected
05/06/12 20:24:56 Client 79.161.16.40 gone
05/06/12 20:24:56 Statistics:
05/06/12 20:24:56 framebuffer updates 0, rectangles 0, bytes 0
05/06/12 20:29:27 Got connection from client 109.230.246.54
05/06/12 20:29:27 Non-standard protocol version 3.4, using 3.3 instead
05/06/12 20:29:28 rfbVncAuthProcessResponse: authentication failed from 109.230.246.54
05/06/12 20:29:28 Client 109.230.246.54 gone
05/06/12 20:29:28 Statistics:
05/06/12 20:29:28 framebuffer updates 0, rectangles 0, bytes 0
This is a problem because eventually tightvnc rejects a new legitimate client session and reports that there were too many authentication failures when the legitimate client tries to do a VNC session. The workaround is to reboot and reloading tightvnc on a frequent basis.