- 1 Sectool
- 1.1 Summary
- 1.2 Owner
- 1.3 Current status
- 1.4 Detailed Description
- 1.5 Benefit to Fedora
- 1.6 Scope
- 1.7 How To Test
- 1.8 User Experience
- 1.9 Dependencies
- 1.10 Contingency Plan
- 1.11 Documentation
- 1.12 Release Notes
- 1.13 Comments and Discussion
The security audit tool and part of an intrusion detection system.
- Targeted release: Fedora 12
- Last updated: (Aug 24 2009)
- Percentage of completion: 99,9%
- RETIRED since Fedora 25
The security tool that can be used both as a security audit as well as a part of an intrusion detection system. It consists of set of tests, library and textual/graphical frontend. Tests are sorted into groups and security levels. Administrators can run selected tests, groups or whole security levels.
Benefit to Fedora
Lower the attack surface by finding possible vulnerabilities, audit the system to find common problems to improve security and performance of scanning system. Encourage users to use Fedora on server and support other distributions based on Fedora like Red Hat Enterprise Linux or CentOS
- Split the current program to backend mechanism and frontend pieces to separate user part and administrator part of application.
- Dbus communication interface for backend
- opotion for not using dbus
- switch communication between backend and frontend to dbus and unix sockets
- PolicyKit integration
- Define policies for PolicyKit
- Improve GUI
- Better user authentication system
How To Test
There is no need for special hardware to test this feature.
install sectool and sectool-gui
$ yum install sectool*
Test the most basic functions to make sure sectool is able to start, etc.
This should show a help for command line user interface
Shows the version of package. Make sure that the version string is correct.
Check that the level numbers are sane. Test globbing:
sectool --list \*home\*
Show information of some tests:
sectool --info <test>
Running the tests
Test that sectool is able to run tests.
- Test running a single test in default level:
sectool --run bootloader
- Test running a single test in non-default level:
sectool --level 5 --run bootloader
Bootloader is a good candidate since it should stay silent in the default level and ERROR in level 5
- Test running multiple levels in default/nondefault mode:
sectool --run bootloader home_files sectool --level 5 --run bootloader home_files
- Test globbing:
sectool --run \*home\*
Running a level
- Run a level:
sectool --level 5
Level 5 might be good if you have enough time, level 1 is probably OK for quick smoke testing.
- Run a level via its nickname:
sectool --level Desktop
The valid levels are: Naive, Desktop, Network, Server, Paranoid.
- Exclude a test from its level:
sectool --level 1 --exclude integrity
Check that all the tests in level 1 except for integrity ran.
- Include a test into another level:
sectool --level 1 --include path
Check that all the tests in level 1 and 'path' ran.
- Check that sectool display additional info when ran with --debug. A good candidate is home_files:
sectool --run home_files --debug
- Check that sectool displays hints when ran with --hint. A good candidate is mountopt:
sectool --run mountopt --hint
Test auto action
- Running a whole level via --action. Set the [ACTION] section of /etc/sectool/sectool.conf as follows:
Run sectool --auto, whole level 1 should run.
- Running some tests on a specific level. Set the [ACTION] section of /etc/sectool/sectool.conf as follows:
[ACTION] LEVEL=5 RUN=path bootloader
Run sectool --auto, tests path and bootloader should run on level 5.
- Running some tests on a default level. Set the [ACTION] section of /etc/sectool/sectool.conf as follows:
[ACTION] RUN=path bootloader
Run sectool --auto, tests path and bootloader should run default level.
- Including and excluding tests from a level in the auto action. Set the [ACTION] section of /etc/sectool/sectool.conf as follows:
[ACTION] LEVEL=1 RUN=+bootloader -integrity
Run sectool --auto, level 1 should run excluding integrity and including bootloader.
Diff and sending a mail
- Check that sectool is able to produce a meaningful diff
sectool --run suid; chmod a+s /bin/cp; sectool --run suid --diff; chmod a-s /bin/cp
- Sending email - vanilla config
that is SEND_BODY=diff SEND_ATTACHMENT=full make sure your sendmail is running and properly configured
sectool --run home_files --mail email@example.com
- Sending email - full output in body, diff in attachment.
Set SEND_ATTACHMENT=diff and SEND_BODY=full in /etc/sectool/sectool.conf
- Sending email via SMTP server
the default is via sendmail configure your server, username and password in /etc/sectool/sectool.conf and change TARGET=smtp
- Check the --clean action
run a test, make sure that results.xml exists
should remove it
- Check that level can be added/removed via config file:
echo "path 1 2" > ~/.sectoolrc
test a level that is not included in the dsc
- Check that disabling all levels work:
echo "path" > ~/.sectoolrc
- See /var/log/sectool.log is created, filled and properly formated
- Check that logrotate handles sectool.log
- Test non-dbus communication by running sectool, e.g.:
sectool -r vsftpd
- Test dbus sectool by running with option --use-dbus:
sectool --use-dbus -r vsftpd
Should end with no authorization message
sectool --use-dbus --list
Should show the list of tests
- Test PolicyKit authentification by running sectool-gui
- Test sectool-gui tool by:
- running tests
- running level
- clear cache data
- stop performing tests
GUI tool should react swiftly without delay, stoping tests should be with approximately 2s delay by stopping all threads.
Users will have option to audit their system to improve security and performance. Users can choose between command line testing through dbus (or without dbus) and GUI testing tool. The whole system will be less vulnerable to attack.
This feature is not critical. Revert to previous release in case that could not be complete by the final feature freeze.
- Switch to using PolicyKit https://bugzilla.redhat.com/show_bug.cgi?id=502733
- Sectool and file capabilities https://bugzilla.redhat.com/show_bug.cgi?id=449990
- Arch dependent files in /usr/share https://bugzilla.redhat.com/show_bug.cgi?id=501825
- Bug in netserv test https://bugzilla.redhat.com/show_bug.cgi?id=503207
- Spurious warnings in filesystem test in mock chroots https://bugzilla.redhat.com/show_bug.cgi?id=503209
- FTBFS sectool-0.9.3-1.fc12 https://bugzilla.redhat.com/show_bug.cgi?id=511473
- Sectool is a security tool that can be used both as a security audit as well as a part of an intrusion detection system. It consists of set of tests, library and textual/graphical frontend. Tests are sorted into groups and security levels. Administrators can run selected tests, groups or whole security levels.
- Sectool is now using PolicyKit and dbus for better security in authentication actions.