Bugzilla Security While some of the items in this chapter are related to the operating system Bugzilla is running on or some of the support software required to run Bugzilla, it is all related to protecting your data. This is not intended to be a comprehensive guide to securing Linux, Apache, MySQL, or any other piece of software mentioned. There is no substitute for active administration and monitoring of a machine. The key to good security is actually right in the middle of the word: U R It. While programmers in general always strive to write secure code, accidents can and do happen. The best approach to security is to always assume that the program you are working with isn't 100% secure and restrict its access to other parts of your machine as much as possible.
Operating System
TCP/IP Ports The TCP/IP standard defines more than 65,000 ports for sending and receiving traffic. Of those, Bugzilla needs exactly one to operate (different configurations and options may require up to 3). You should audit your server and make sure that you aren't listening on any ports you don't need to be. It's also highly recommended that the server Bugzilla resides on, along with any other machines you administer, be placed behind some kind of firewall.
System User Accounts Many daemons, such as Apache's httpd or MySQL's mysqld, run as either root or nobody. This is even worse on Windows machines where the majority of services run as SYSTEM. While running as root or SYSTEM introduces obvious security concerns, the problems introduced by running everything as nobody may not be so obvious. Basically, if you run every daemon as nobody and one of them gets compromised it can compromise every other daemon running as nobody on your machine. For this reason, it is recommended that you create a user account for each daemon. You will need to set the option in localconfig to the group your web server runs as. This will allow ./checksetup.pl to set file permissions on Unix systems so that nothing is world-writable.
The <filename>chroot</filename> Jail If your system supports it, you may wish to consider running Bugzilla inside of a chroot jail. This option provides unprecedented security by restricting anything running inside the jail from accessing any information outside of it. If you wish to use this option, please consult the documentation that came with your system.
MySQL
The MySQL System Account As mentioned in , the MySQL daemon should run as a non-privileged, unique user. Be sure to consult the MySQL documentation or the documentation that came with your system for instructions.
The MySQL <quote>root</quote> and <quote>anonymous</quote> Users By default, MySQL comes with a root user with a blank password and an anonymous user, also with a blank password. In order to protect your data, the root user should be given a password and the anonymous user should be disabled. Assigning the MySQL <quote>root</quote> User a Password bash$ mysql mysql mysql> UPDATE user SET password = password('new_password') WHERE user = 'root'; mysql> FLUSH PRIVILEGES; Disabling the MySQL <quote>anonymous</quote> User bash$ mysql -u root -p mysql Enter Password: new_password mysql> DELETE FROM user WHERE user = ''; mysql> FLUSH PRIVILEGES; This command assumes that you have already completed .
Network Access If MySQL and your web server both run on the same machine and you have no other reason to access MySQL remotely, then you should disable the network access. This, along with the suggestion in , will help protect your system from any remote vulnerabilities in MySQL. Disabling Networking in MySQL Simply enter the following in /etc/my.cnf: [mysqld] # Prevent network access to MySQL. skip-networking
Web server
Disabling Remote Access to Bugzilla Configuration Files There are many files that are placed in the Bugzilla directory area that should not be accessible from the web server. Because of the way Bugzilla is currently layed out, the list of what should and should not be accessible is rather complicated. A quick way is to run testserver.pl to check if your web server serves Bugzilla files as expected. If not, you may want to follow the few steps below. Bugzilla ships with the ability to create .htaccess files that enforce these rules. Instructions for enabling these directives in Apache can be found in In the main Bugzilla directory, you should: Block: *.pl *localconfig* In data: Block everything But allow: duplicates.rdf In data/webdot: If you use a remote webdot server: Block everything But allow *.dot only for the remote webdot server Otherwise, if you use a local GraphViz: Block everything But allow: *.png *.gif *.jpg *.map And if you don't use any dot: Block everything In Bugzilla: Block everything In template: Block everything Be sure to test that data that should not be accessed remotely is properly blocked. Of particular interest is the localconfig file which contains your database password. Also, be aware that many editors create temporary and backup files in the working directory and that those should also not be accessible. For more information, see bug 186383 or Bugtraq ID 6501. To test, simply run testserver.pl, as said above. Be sure to check for instructions specific to the web server you use.
Bugzilla
Prevent users injecting malicious Javascript If you installed Bugzilla version 2.22 or later from scratch, then the utf8 parameter is switched on by default. This makes Bugzilla explicitly set the character encoding, following a CERT advisory recommending exactly this. The following therefore does not apply to you; just keep utf8 turned on. If you've upgraded from an older version, then it may be possible for a Bugzilla user to take advantage of character set encoding ambiguities to inject HTML into Bugzilla comments. This could include malicious scripts. This is because due to internationalization concerns, we are unable to turn the utf8 parameter on by default for upgraded installations. Turning it on manually will prevent this problem.