Open Stack

Timur:
Thanks. Yes. That’s part of the plan.


Thanks and have a great day.

Yours,
Michael

-----------------------------------------------------------------------------
Michael Xin | Manager, Security Engineering - US
Product Security  |Rackspace Hosting
Office #: 501-7341   or  210-312-7341
Mobile #: 210-284-8674
5000 Walzem Road, San Antonio, Tx 78218
----------------------------------------------------------------------------
Experience fanatical support


From: Timur Nurlygayanov <tnurlygayanov at mirantis.com<mailto:tnurlygayanov at mirantis.com>>
Date: Thursday, June 18, 2015 at 11:23 AM
To: michael xin <michael.xin at rackspace.com<mailto:michael.xin at rackspace.com>>
Cc: Charles Neill <charles.neill at RACKSPACE.COM<mailto:charles.neill at RACKSPACE.COM>>, "Clark, Robert Graham" <robert.clark at hp.com<mailto:robert.clark at hp.com>>, "openstack-security at lists.openstack.org<mailto:openstack-security at lists.openstack.org>" <openstack-security at lists.openstack.org<mailto:openstack-security at lists.openstack.org>>
Subject: Re: [Openstack-security] [security] [QA] Do we have security tests suite for OpenStack components?

Thank you all for the responses!

We are planing to contribute our security tests too and we are looking for the right place for this.

Michael, do you plan to add some security tests to Tempest as well?

On Thu, Jun 18, 2015 at 5:33 PM, Michael Xin <michael.xin at rackspace.com<mailto:michael.xin at rackspace.com>> wrote:
Charles and Clark:
Thanks for the update. There are two different types of fuzzing that are discussed here:

  1.  Deep security fuzzing to find new security defects. Sulley looks like a good tool for this purpose. For this type of fuzzing, the goal is to generate huge amount of unexpected fuzz attack vectors to check whether the application could handle these data. It will take a long time and it has the potential to find new security defects.
  2.  Light fuzzing automation that can be integrated with CI process. For this type of fuzzing, we have pre-defined fuzzing strings and we know the expected responses back. We can have a list of security tests defined: authentication & authorization, injection issues, security transport layer issues, user input validation, denial of services, etc. It should give us consistent level of security coverage. This approach has the risk of missing new security defects. But once new security defects are identified, it is possible to add security checks to make sure that they will be identified in the future. This types of security tests can be run the same time that QA function tests are run. It is easy for integrate. Rackspace has been working with our QE to build more security test cases like these. For OpenStack, we are currently pushing security test cases to Barbican and Poppy projects. More should come in the future.


Thanks and have a great day.

Yours,
Michael

-----------------------------------------------------------------------------
Michael Xin | Manager, Security Engineering - US
Product Security  |Rackspace Hosting
Office #: 501-7341   or  210-312-7341
Mobile #: 210-284-8674
5000 Walzem Road, San Antonio, Tx 78218
----------------------------------------------------------------------------
Experience fanatical support


From: Charles Neill <charles.neill at RACKSPACE.COM<mailto:charles.neill at RACKSPACE.COM>>
Date: Wednesday, June 17, 2015 at 10:57 AM
To: "Clark, Robert Graham" <robert.clark at hp.com<mailto:robert.clark at hp.com>>, Timur Nurlygayanov <tnurlygayanov at mirantis.com<mailto:tnurlygayanov at mirantis.com>>, "openstack-security at lists.openstack.org<mailto:openstack-security at lists.openstack.org>" <openstack-security at lists.openstack.org<mailto:openstack-security at lists.openstack.org>>

Subject: Re: [Openstack-security] [security] [QA] Do we have security tests suite for OpenStack components?

I'm actually working on some (very simple) fuzzing tests for the Barbican project right now. I've realized that using a Python client that is strictly defined for sending "good" data to an API isn't necessarily the best mechanism for truly fuzzing that API. For example, the "requests" lib chokes on trying to convert unicode strings to ascii in HTTP request headers.

I'm contemplating using something like sulley (link: https://github.com/OpenRCE/sulley) for deeper fuzzing, but I don't have time to really dig into it at the moment. I'd love to hear others' thoughts on this problem.

Cheers,
Charles Neill

From: <Clark>, Robert Graham <robert.clark at hp.com<mailto:robert.clark at hp.com>>
Date: Wednesday, June 17, 2015 at 5:32 AM
To: Timur Nurlygayanov <tnurlygayanov at mirantis.com<mailto:tnurlygayanov at mirantis.com>>, "openstack-security at lists.openstack.org<mailto:openstack-security at lists.openstack.org>" <openstack-security at lists.openstack.org<mailto:openstack-security at lists.openstack.org>>
Subject: Re: [Openstack-security] [security] [QA] Do we have security tests suite for OpenStack components?

Fuzzing with Tempest is something that was discussed at the last summit, I think Rackspace had some interesting work that wanted to share but I might be miss-remembering.

One of the issues is that unless you’re looking with issues in the python libraries used for handling the APIs (which would be interesting to find) message/protocol fuzzing is pretty limited, instead we need something that really understands the API to perform fuzzing to find things that logically shouldn’t be allowed to happen – many of these will be broken state issues.

-Rob

From: Timur Nurlygayanov [mailto:tnurlygayanov at mirantis.com]
Sent: 15 June 2015 18:21
To: openstack-security at lists.openstack.org<mailto:openstack-security at lists.openstack.org>
Subject: [Openstack-security] [security] [QA] Do we have security tests suite for OpenStack components?

Hi team,
Looks like we are using Bandit framework [1] for static analysis of code of different OpenStack components, but I can't find some integration security tests for OpenStack components. Do we have some additional open-source test framework / tests suite for security testing of OpenStack components?
I found the blueprint in Tempest about fuzzy testing [2], so, we can start to develop some security tests in Tempest and use them to perform security testing on the integration level and also to validate some security bug fixes.
Do we have some list with scenarios, which we can cover with fuzzing tests in Tempest? We can start from some tests which will validate fixed security issues, it will be really helpful if you can share some ideas about tests, which we have to create.

Thank you!

[1] https://github.com/stackforge/bandit
[2] https://blueprints.launchpad.net/tempest/+spec/fuzzy-test

--

Timur,
Senior QA Engineer
OpenStack Projects
Mirantis Inc



--

Timur,
Senior QA Engineer
OpenStack Projects
Mirantis Inc
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-security/attachments/20150618/4e8defac/attachment.html>