Compliance, Integrity Checks & Audit Evidence
Key Takeaways
Use the vendor advisory and exact installed package build to confirm whether a CVE fix has been backported.
An AIDE comparison uses a trusted file baseline. An RPM signature check verifies a package artifact, while RPM installed-file verification compares against packaged metadata.
Apply a SCAP profile for the target release and record results and exceptions with the control evidence.
File overwrites depend on storage behavior. Match the sanitization method to the medium and verify its result.
Compliance, Integrity Checks & Audit Evidence
A compliance check asks whether a host meets a stated control and whether you can produce evidence for it. A vulnerability check asks whether a known exposure applies to the installed software and configuration. A file-integrity check asks what has changed since a trusted baseline. Each has a different reference point. Record which one answered the question before labeling a host compliant, vulnerable, or compromised.
Decide what an alert proves
An indicator of compromise (IOC) is evidence worth investigating, such as an unexpected executable, a suspicious authentication record or a match to a known malicious file digest. Keep the alert, time, host and data source together. Anti-malware and rootkit-checking tools such as rkhunter generate leads. Compare findings with known changes and other evidence before deciding how to contain or investigate the host. An approved update can change a file hash. An unexplained change to an authentication program deserves closer scrutiny.
A Common Vulnerabilities and Exposures (CVE) identifier describes a publicly recorded vulnerability. The Common Vulnerability Scoring System (CVSS) captures severity characteristics. Include your host’s exposure and the vendor’s advisory when prioritizing action, since a score alone cannot establish that a particular host is affected. Use approved port inventories and protocol inspection to understand which services are exposed and how they behave. Review service configuration as well as software versions.
Worked example: a version-only scanner finding
A scan flags a library because the upstream version appears older than the release that fixed a CVE. On a Red Hat system, record the exact installed RPM release with rpm -q PACKAGE and check the advisory for that operating-system release and package build. A vendor can backport the security fix to an older upstream version. If the advisory marks the installed build fixed, record that advisory and the installed package identity when resolving the scanner finding. If the host actually lacks the vendor’s fix, apply the approved update and recheck the installed build. A version string without its vendor package release leaves the question open.
Match the integrity tool to the question
| Question | Evidence to collect |
|---|---|
| Has a file changed since a known-good state? | An Advanced Intrusion Detection Environment (AIDE) baseline can record hashes and attributes. Initialize it on a trusted system, protect a copy, and use aide --check to compare later. Investigate differences before accepting a new baseline. |
| Is this downloaded RPM signed by a trusted key and intact? | Use rpmkeys --checksig PACKAGE.rpm and review signature results against the trusted vendor key. Verify the origin of the key separately. |
| Have files installed by an RPM changed since installation? | Use rpm -V PACKAGE and examine differences against the package database. A deliberate administrator change to configuration can account for a difference. |
| Does the host meet a policy profile? | Use a Security Content Automation Protocol (SCAP) profile applicable to the distribution and release. OpenSCAP can evaluate a profile and produce a report. Review failed and inapplicable checks. |
A benchmark such as a Center for Internet Security (CIS) Benchmark specifies recommended checks for a given platform. Document the version, profile, host scope, exceptions and supporting results. For a software supply-chain control, collect package provenance, configured repositories and signing keys. A signature gives evidence about integrity and the signer under the configured trust policy. A signed package still needs vulnerability and configuration review.
Sanitization and banners
Select a sanitization method that matches the medium and policy. shred assumes data is overwritten in place. Copy-on-write, journaling, snapshots and SSD wear leveling can leave copies outside that path. An overwrite with dd if=/dev/urandom uses random bytes as input and reaches the addressed device blocks, while badblocks -w is a destructive surface test. Neither is a blanket substitute for a validated sanitization method. Cryptographic erase depends on the prior encryption and protection of the data and the successful sanitization of applicable keys. Account for retained key copies, backups and snapshots when they are in scope. Record the device, method, result and verification.
A local text login commonly reads a notice from /etc/issue. A service may use /etc/issue.net as its remote pre-login banner when explicitly configured; OpenSSH uses the file named by its Banner directive. /etc/motd is commonly shown after successful login when the service is configured to display it. Test from the relevant console or client and record what the intended user sees.
A scan flags a supported Red Hat package based only on an older upstream version string. Which evidence can settle whether the distributor’s fix is installed?
The name of the repository alone
The size of the RPM file
The exact installed package release and the vendor advisory for that operating-system version
The process ID of the application
You have an AIDE baseline created from a known-good host. Which check answers whether a monitored configuration file has changed since then?
Read its current open file descriptor count
Check a newly downloaded RPM’s signature
Run a port inventory
Run aide --check and review the differences from the baseline
A lab host stores sensitive data on an SSD with copy-on-write snapshots. After a successful file-level overwrite, what evidence is still required to close a sanitization control?
Only the file modification time
A validated method for the medium and evidence that in-scope copies were addressed
Only the exit status from the overwrite
The result of a port scan
Sections you finish are checked off in the contents.