Server logs contain information that cannot be found anywhere else. For example, server errors and user access records.


Having the ability to review these logs gives the ability to determine what caused an issue on the server.



  • Make sure logs are working properly.

  • syslog on Debian contains everything except authentication-related messages.

  • Authentication related events can be found in auth.log. Use it to investigate failed login attempts, brute-force attacks and other vulnerabilities related to user authorisation mechanisms. RedHat and CentOS based systems use the secure file for tracking all security related messages including authentication failures, sudo logins, ssh logins and other errors logged by the system security services daemon and is very useful for detecting hacking attempts.

  • Additional logs that may give clues are

    • faillog containing information on failed login attempts and can be used to detect attempted security breaches involving username/password hacking and brute-force attacks.

  • Normally I am all for distributed and decentralised systems, but for an infrastructure with multiple servers, centralised logging saves a lot of time and energy.

  • When setting up a machine to act as log server, harden it and only use it to collect logs from other machines on your network. It should not be running anything else.

  • Log what is needed to track the path of a common intruder.

  • If and when it is all running well, consider setting up for intrusion detection and log correlation.