Time to discuss OVAL (Open Vulnerability Assessment Language). In all the previous posts I focused the checking of rules (does the system comply with the given rule) on scripts, through the Script Check Engine supported by openscap. The advantage of SCE is that most people can quickly provide automated checks to run in script format. But SCE has a few downsides.
- You cannot guarantee that the scripts will do no harm on the system. A badly written script might manipulate system settings, get a huge amount of resources, leave stale result files on the system, flood file systems and more. If you get scripts from other parties, you'll need to review them thoroughly before running them against all your systems. Especially when you run the compliance validation tool (openscap in our example) as root.
- SCE support is only available for openscap (and perhaps one or two others) as it is not an international standard. If you use any of the SCAP validated tools then you will not be able to benefit from the SCE scripts. And that would make the XCCDF document back to a purely documenting best practice.
- Every rule requires separate scripts, even though many of the rules will be very similar and thus reuse a lot of the scripts.
OVAL on the other hand provides those advantages. An OVAL file is an XML file that contains the tests to run, in an (I must say) somewhat complex manner. Really, OVAL is not simple, but it does contain advantages that SCE doesn't.
- It is a standard, part of the SCAP standards. OVAL files are reusable across multiple tools, allowing you to focus once on the rules rather than having to rewrite the rules for every time you change the tool.
- OVAL can be platform-agnostic. Of course, not all tests are platform-agnostic (validating registry keys is a Windows-only check) but many are.
- All rules can be mentioned in a single file (or spread across
multiple files if that makes management easier), but more
importantly rules will also reuse definitions from other rules. If
you have three rules that pertain to a file (say
/etc/rc.conf) then the definition of that file is shared across all rules.
- The OVAL standard is designed to be non-intrusive. All declarations you do in an OVAL file are pure read-only statements. This gives more confidence to have OVAL statements from third parties ran across your organization. Of course, reviewing them never hurts, but you already know that they will not modify any setting.
Like SCE, OVAL checks are individual checks that are executed and returned. They too return a success or failure (or error) and can deliver more detailed information as part of their result (like the SCE results output we looked at before) so allow administrators to investigate further why a rule failed (without needing to log on to the system and look for themselves).
A basic structure of OVAL is a definition that describes what the rule is for. The definition refers to one or more tests that are evaluated on a system. These tests refer to an object that needs to be checked, and optionally a state to which the object should (or shouldn't) match.
Consider the test we made with SCE to see if a platform is a Gentoo Linux system:
1 2 3 4 5
In OVAL, this would be structured as follows (pseudo-OVAL):
- The system is a Gentoo Linux system
- The object that represents /etc/gentoo-release must exist
- The /etc/gentoo-release file
The resulting OVAL file is quite complex for a simple rule. I see the
OVAL complexity as part of a normalization (similar to database
normalization) process to allow higher reuse. If we later want to check
the content of the
gentoo-release file, we will reuse the definition
(object with id oval:org.gentoo.dev.swift:obj:1) rather than making a
second object for it, and use that definition to create new tests.
The structure of OVAL is the same everywhere. First define the
definitions, then the tests, then the objects and then, optionally, the
states. A very important aspect is to have the identifiers (
correct. The structure of OVAL identifiers is standardized as well:
(XML content lost during blog conversion)
- Like the namespace used in XCCDF documents, this is the reverse notation of a domainname. In the example above, this is org.gentoo.dev.swift.
- The type of the entry in OVAL. This can be def (definition), tst (test), obj (object), ste (state) or var (variable).
- The identifier of this particular entry. This identifier has to be a positive integer.
By standardizing the identifiers, you can create repositories within
your organization, and have other teams reuse your OVAL components when
needed. As the identifier remains the same (even when you update the
OVAL object itself to be more precise) those tests keep validating
correctly. For instance, say that Gentoo Linux would be changed in the
future not to provide a
gentoo-release file anymore, but
gentoo-linux-release file instead (not that it is planning that, it is
just hypothetical), then you can update the test (with description
"Gentoo Linux is installed") to check if either of the two files exist:
(XML content lost due to blog conversion)
If we save all Gentoo releated OVAL statements in a file called
gentoo-oval.xml then we can update the
gentoo-cpe.xml file (which we
discussed in the past) to the following:
(XML content lost during blog conversion)
With this change, openscap (or any other XCCDF interpreter) will use the
OVAL definition to see if a platform is Gentoo Linux or not, and does
not need to execute the
gentoo-platform.sh script anymore, which is
now fully deprecated and superceded by the OVAL statement.
In the next posts, I'll write up one of the other tests we had (which checks the content of a file - one of the most used tests I think) in OVAL, and have the XCCDF document updated to only use OVAL statements.
This post is part of a series on SCAP content: