NIST 800-53 compliance
DETAILS: Tier: Ultimate Offering: Self-managed, GitLab Dedicated
This page provides a reference for GitLab administrators who want to configure self-managed instances to meet applicable NIST 800-53 controls. GitLab does not provide specific configuration guidance because of the variety of requirements an administrator might have. Before you deploy a GitLab instance that meets the NIST 800-53 security controls, you should work with a customer solutions architect for technical details.
Scope
This page follows the structure of the NIST 800-53 control families. Because the scope of the page is limited primarily to configurations made to GitLab itself, not all control families apply. Configuration details are intended to be infrastructure agnostic.
GitLab guidance does not constitute a fully compliant system. Before you handle government data, you should:
- Plan for additional configuration and hardening of your entire technology stack.
- Consider an independent assessment of security configurations.
- Understand the differences in deployments across supported cloud providers and follow specific guidance where available.
Compliance features
GitLab offers several compliance features you can use to automate critical controls and workflows in GitLab. Before you make configurations aligned with NIST 800-53, you should enable these foundational features.
Configuration by control family
System and Service Acquisition (SA)
GitLab is a DevSecOps platform that integrates security throughout the development lifecycle. At its core, you can use GitLab to address a wide range of controls within the SA control family.
System development lifecycle
You can use GitLab to meet the core of this requirement. GitLab provides a platform where work can be organized, planned, and tracked. NIST 800-53 requires that security is incorporated into the development of the application. You can configure CI/CD pipelines to continuously test code while it ships and simultaneously enforce security policies. GitLab includes a suite of security tools that you can incorporate into the development of customer applications, including but not limited to:
- Security configuration
- Container Scanning
- Dependency Scanning
- Static Application Security Testing
- Infrastructure as Code (IaC) Scanning
- Secret Detection
- Dynamic Application Security Testing (DAST)
- API fuzzing
- Coverage-guided fuzz testing
Beyond the CI/CD pipeline, GitLab provides detailed guidance on how to configure releases. Releases can be created with a CI/CD pipeline and take a snapshot of any branch of source code in a repository. Instructions for creating releases are included in Create a Release. An important consideration for NIST 800-53 or FedRAMP compliance is that released code may need to be signed to verify authenticity of code and satisfy requirements in the System and Information Integrity (SI) control family.
Access Control (AC) and Identification and Authentication (IA)
Access management in a GitLab deployment is unique to each customer. GitLab provides a range of documentation that covers deployments with identity providers and GitLab native authentication configurations. It is important to consider the organizational requirements prior to determining how to approach authentication to a GitLab instance.
Identity Providers
Access within GitLab can be managed with the UI or by integrating with an existing identity provider. In order to meet FedRAMP requirements, ensure that the existing identity provider is FedRAMP authorized on the FedRAMP Marketplace. To meet requirements such as PIV, you should leverage an identity provider rather than using native authentication in self-managed GitLab.
GitLab provides resources for configuring various identity providers and protocols, including
-
More information on identity providers can be found in GitLab Docs.
Native GitLab User Authentication Configurations
Account management and classification - GitLab empowers administrators to keep track of users with varying degrees of sensitivity and access requirements. GitLab supports the concept of least privilege and role based access by providing options for granular access. At the project level, the following roles are supported
-
Guest
-
Reporter
-
Developer
-
Maintainer
-
Owner
Additional details on project level permissions can be found in the documentation. GitLab also supports custom roles for customers that have unique permission requirements.
GitLab also supports the following user types for unique use cases:
-
Auditor Users - The auditor role provides read-only access to all groups, projects and other resources except for the Admin area and project/group settings. You can use the auditor role when engaging with third-party auditors that require access to certain projects to validate processes.
-
External Users - External users can be set to provide limited access for users that may not be part of the organization. Typically, this can be used to satisfy managing access for contractors or other third parties. Controls such as IA-4(4) require non-organizational users to be identified and managed in accordance with company policy. Setting external users can reduce risk to an organization by limiting access to projects by default and assisting administrators in identifying which users are not employed by the organization.
-
Service Accounts - Service accounts may be added to accommodate automated tasks. Service accounts do not use a seat under the license.
Admin area - In the Admin area, administrators can export permissions, review user identities, administer groups, and much more. Functions that can be used to meet FedRAMP / NIST 800-53 requirements:
-
Reset user password when suspected of compromise.
-
Unlock users. By default, GitLab locks users after 10 failed sign-in attempts. Users remain locked for 10 minutes or until an administrator unlocks the user. In GitLab 16.5 and later, administrators can use the API to configure max login attempts and time period for remaining locked out. Per guidance in AC-7, FedRAMP defers to NIST 800-63B for defining parameters for account lockouts, which the default setting satisfies.
-
Review abuse reports or spam logs. FedRAMP requires organizations to monitor accounts for atypical use (AC-2(12)). GitLab empowers users to flag abuse in abuse reports, where administrators can remove access pending investigation. Spam logs are consolidated in the Spam logs section of the Admin area. Administrators can remove, block, or trust users flagged in that area.
-
Set password storage parameters. Stored secrets must satisfy FIPS 140-2 or 140-3 as outlined in SC-13. PBKDF2+SHA512 is supported with FIPS compliant ciphers when FIPS mode is enabled.
-
Credentials inventory enables administrators to review all secrets used in a GitLab self-managed instance in one place. A consolidated view of credentials, tokens, and keys may assist with satisfying requirements such as reviewing passwords or rotating credentials.
-
Set customer password length limits. FedRAMP defers to NIST 800-63B in IA-5 for establishing password length requirements. GitLab supports 8-128 character passwords, with 8 characters set as the default. GitLab provides instructions for updating the minimum password length with the GitLab UI, which organizations interested in enforcing longer passwords can use. Additionally, self-managed customers may configure complexity requirements through the Admin area UI.
-
Default session durations - FedRAMP establishes that users that have been inactive for a set time period should be logged out. FedRAMP does not specify the time period, however, clarifies that for privileged users they should be logged out at the end of the standard work period. Administrators can establish default session durations.
-
Provisioning New Users - Administrators can create new users for their GitLab account with the Admin area UI. In compliance with IA-5, GitLab requires new users to change their passwords on first login.
-
Deprovisioning Users - Administrators are able to remove users with the Admin area UI. An alternative to deleting users is to block a user and remove all access. Blocking a user maintains their data in repositories while removing all access. Blocked users do not impact seat counts.
-
Deactivate Users - Inactive users that have been identified during account reviews may be temporarily deactivated. Deactivation is similar to blocking, but there are a few important differences. Deactivating a user does not prohibit the user from signing into the GitLab UI. A deactivated user can become active again by signing in. A deactivated user:
-
Cannot access repositories or the API.
-
Cannot use slash commands. For more information, see slash commands.
-
Does not occupy a seat.
-
Additional Identification Methods
Two-factor authentication - GitLab supports the following second factors:
-
One-time password authenticators
-
WebAuthn devices
Instructions for enabling two-factor authentication are provided in the documentation. Customers pursuing FedRAMP must consider two-factor providers that are FedRAMP authorized and support FIPS requirements. FedRAMP authorized providers can be found on the FedRAMP Marketplace. When selecting a second factor, it is important to note that NIST and FedRAMP are now indicating that phishing resistant authentication, such as WebAuthn, must be used (IA-2).
SSH keys
-
GitLab provides instructions on how to configure SSH keys to authenticate and communicate with Git. Commits can be signed, providing additional verification for anyone with a public key.
-
Keys should be configured to meet applicable strength and complexity requirements, such as using FIPS 140-2 and FIPS 140-3 validated ciphers . Administrators can restrict minimum key technologies and key lengths. Additionally, administrators can block or ban compromised keys.
Personal access tokens
Personal access tokens for user access are disabled by default in FIPS enabled instances.
Other Access Control Family Concepts
System Use Notifications
Federal requirements often outline the need for a banner at login. This can be configured through an identity provider and through the GitLab banner functionality.
External Connections
It is important to document all external connections and ensure that they meet compliance requirements. For example, setting up an API integration with a third party may violate data handling requirements, depending on how that third party secures customer data. It is important to review all external connections and understand their security impacts prior to enabling them. For customers pursuing FedRAMP or similar certifications, connecting to other non-FedRAMP authorized services or services of a lower data impact level may violate the authorization boundary.
Personal Identity Verification (PIV)
Personal Identification Verification cards may be a requirement for organizations meeting federal requirements. In order to meet PIV requirements, GitLab requires customers to connect PIV-enabled identity solutions with SAML. A link to SAML documentation is provided earlier in this guide.
Audit and Accountability (AU)
NIST 800-53 requires organizations to monitor for security relevant events, analyze those events, generate alerts, and investigate alerts in accordance with the criticality of the alerts. GitLab provides a wide array of security events for monitoring that can be routed to a Security Information and Event Management (SIEM) solution.
Event Types
GitLab outlines the configurable audit event log types, which can be streamed and/or saved to a database. Administrators are able to configure the events that they'd like captured for their GitLab instance.
Log System
GitLab includes an advanced log system where everything can be logged. GitLab offers guidance on log system log types, which include a wide range of outputs. Review the linked guidance for further details.
Streaming Events
GitLab administrators can stream audit events to a SIEM or other storage location using the event streaming functionality. Administrators can configure multiple destinations and set event headers. GitLab provides examples for event streaming which outline headers, payloads for HTTP and HTTPS events, and much more.
It is important for administrators to review the FedRAMP or NIST 800-53 AU-2 requirements and implement audit events that map to the required audit event type. AU-2 identifies the following event buckets:
-
Successful and unsuccessful account logon events
-
Account management events
-
Object access
-
Policy change
-
Privilege functions
-
Process tracking
-
System events
-
For Web applications:
-
All administrator activity
-
Authentication checks
-
Authorization checks
-
Data deletions
-
Data access
-
Data changes
-
Permission changes
-
Administrators should consider both the required event types and any additional organizational requirements when enabling events in GitLab.
Metrics
Outside of security events, administrators may also want visibility into the performance of their application to support uptime. GitLab provides a robust set of documentation around metrics that are supported in a GitLab instance.
Storage
Customers are responsible for ensuring that logs are stored in a long-term storage solution that meets compliance requirements. FedRAMP, for example, requires logs to be stored for 1 year. Customer organizations may also need to meet National Archives and Records Administration requirements, depending on the impact of the collected data. It is important to review the impact of records collected and understand the applicable compliance requirements.
Incident Response (IR)
Once audit events have been configured, those events must be monitored. GitLab provides a centralized management interface for compiling system alerts from a SIEM or other security tooling, triaging alerts and incidents, and informing stakeholders. The incident management documentation outlines how GitLab can be used to run the aforementioned activities in a security incident response organization.
Incident Response Lifecycle
GitLab can manage the entirety of the incident response lifecycle for an organization. Review the following resources, which may help meet incident response requirements:
Configuration Management (CM)
Change Control
GitLab, at its core, can satisfy configuration management requirements related to change control. Issues and merge requests are the primary methods for supporting changes.
Issues are a flexible platform for capturing metadata and approvals prior to implementing changes. Consider reviewing GitLab documentation on planning and tracking work to gain a full understanding of how GitLab features can be used to satisfy configuration management controls.
Merge requests offer a method for standardizing changes from a source branch to a target branch. In the context of NIST 800-53, it is important to consider how approvals should be collected prior to merging code and who has the ability to merge code within the organization. GitLab provides guidance on the various settings available for approvals in merge requests. Consider assigning approval and merge privileges only to appropriate roles after the necessary reviews have been completed. Additional merge settings to consider:
-
Remove all approvals when a commit is added - Ensures that approvals are not carried over when new commits are made to a merge request.
-
Restrict individuals who can dismiss code change reviews.
-
Assign code owners to be notified when sensitive code or configurations are changed through merge requests.
-
Ensure all open comments are resolved before allowing code change merging.
-
Configure push rules - Push rules can be configured to meet requirements such as reviewing signed code, verifying users, and more.
Testing and Validation of Changes
CI/CD pipelines are a critical component of testing and validating changes. It is the responsibility of the customer to implement sufficient testing and validation pipelines for specific use cases. When selecting services, consider where that pipeline will run. Connecting to external services may violate an established authorization boundary where federal data is permitted to be stored and processed. GitLab provides runner container images configured to run on FIPS-enabled systems. GitLab provides hardening guidance for pipelines, including how to configure protected branches and implement pipeline security. Additionally, customers may want to consider assigning required checks before merging code to ensure that all checks have passed prior to updating the code.
Component Inventory
NIST 800-53 requires cloud services providers to maintain component inventories. GitLab cannot directly track underlying hardware, however, it can generate software inventories via container and dependency scanning. GitLab outlines the dependencies that container scanning and dependency scanning can detect. GitLab offers additional documentation around generating dependency lists, which can be used in software component inventories. Software Bill of Materials support is covered further down in this document, under Supply Chain Risk Management.
Container Registry
GitLab provides an integrated container registry to store container images for GitLab projects, which can be used as an authoritative repository for deploying containers in a highly virtualized and scalable environment. Container registry administration guidance is available for review.
Contingency Planning (CP)
GitLab provides guidance and services that can help meet the core contingency planning requirements. It is important to review the included documentation and plan accordingly to meet organizational requirements for contingency planning activities. Contingency planning is unique to each organization so it is important to consider organizational needs prior to establishing a contingency plan.
Selecting a GitLab Architecture
GitLab provides extensive documentation on the architectures supported in a self-managed instance. GitLab supports the following cloud service providers:
GitLab provides a decision tree for assisting customers with selecting reference architectures and availability models. Most cloud service providers provide resiliency in a region for managed services. When selecting an architecture, it is important to consider the organization's tolerance for downtime and the criticality of data. GitLab Geo can be considered for additional replication and failover capabilities.
Identify Critical Assets
NIST 800-53 requires the identification of critical assets to ensure their prioritized restoration during an outage. Critical assets to consider include Gitaly nodes and PostgreSQL databases. Customers should identify additional assets that need backups or replication as appropriate.
Backups
GitLab Docs outlines backup strategies for critical components, including:
GitLab Geo
GitLab Geo is likely to be a critical component of any implementation pursuing compliance with NIST 800-53. It is important to review the available documentation to ensure that Geo is configured appropriately for each use case.
Implementing Geo provides the following benefits:
-
Reduce from minutes to seconds the time taken for distributed developers to clone and fetch large repositories and projects.
-
Enable developers to contribute ideas and work in parallel, across regions.
-
Balance the read-only load between primary and secondary sites.
-
Can be used for cloning and fetching projects, in addition to reading any data available in the GitLab web interface (see limitations).
-
Overcomes slow connections between distant offices, saving time by improving speed for distributed teams.
-
Helps reduce the loading time for automated tasks, custom integrations, and internal workflows.
-
Can quickly fail over to a secondary site in a disaster recovery scenario.
-
Allows planned failover to a secondary site.
Geo provides the following core features:
-
Read-only secondary sites: Maintain one primary GitLab site while still enabling read-only secondary sites for distributed teams.
-
Authentication system hooks: Secondary sites receive all authentication data (like user accounts and logins) from the primary instance.
-
An intuitive UI: Secondary sites use the same web interface as the primary site. In addition, there are visual notifications that block write operations and make it clear that a user is in a secondary site.
Additional Geo Resources:
PostgreSQL
GitLab provides guidance on how to configure PostgreSQL clusters with replication and failover. Depending on the criticality of data and maximum tolerable downtime for the GitLab instance, consider configuring PostgreSQL with replication and failover enabled.
Gitaly
When configuring Gitaly, consider the tradeoffs between availability, recoverability, and resilience. GitLab provides extensive documentation on Gitaly capabilities that should assist with determining the correct configuration to meet NIST 800-53 requirements.
Planning (PL)
The planning control family includes maintenance of policies, procedures and other controlled documents. Consider leveraging GitLab to manage the lifecycle of controlled documents. For example, controlled documents can be stored in Markdown as a version-controlled state. Any changes to documents must be made through merge requests, which enforce an organization's approval rules. Merge requests provide a clear history of changes made to a controlled document, which you can use during an audit to demonstrate annual reviews and approvals by appropriate personnel, such as document owners.
Risk Assessment and System and Information Integrity (RA)
Scanning
NIST 800-53 requires continuous monitoring for vulnerabilities and flaw remediation. In addition to infrastructure scanning, compliance frameworks like FedRAMP have scoped in containers and DAST scans into monthly reporting requirements. GitLab provides security tooling that can support container scanning, including Trivy and Grype scanners. Additionally, GitLab provides dependency scanning functionality. Dynamic Application Security Testing (DAST) in GitLab can be used to satisfy web application scanning requirements. GitLab DAST can be configured to run in a pipeline and produce vulnerability reports for running web applications.
Additional security features that may be used to secure and manage application code include:
Patch Management
GitLab documents its Release and Maintenance Policy in the documentation. Prior to upgrading a GitLab instance, please review the available guidance, which can assist with planning an upgrade, upgrading without downtime, and other upgrade paths.
Security dashboards can be configured to track vulnerability data over time, which you can use to identify trends in vulnerability management programs.
Supply Chain Risk Management (SR)
Software Bill of Materials
GitLab dependency and container scanners support the generation of SBOMs. Enabling SBOM reports in container and dependency scanning can empower customer organizations to understand their software supply chain and the inherent risks associated with software components. GitLab scanners support CycloneDX formatted reports.
System and Communication Protection (SC)
FIPS Compliance
Compliance programs based on NIST 800-53, such as FedRAMP, require FIPS compliance for all applicable cryptographic modules. GitLab has released FIPS versions of its container images and provides guidance on how to configure GitLab to meet FIPS compliance standards. It is important to note that certain features are not available or supported in FIPS mode.
While GitLab provides FIPS-compliant images, it is the responsibility of the customer to configure underlying infrastructure and evaluate the environment to confirm FIPS-validated ciphers are enforced.
System and Information Integrity (SI)
Security Alerts, Advisories, and Directives
GitLab maintains an advisory database for tracking security vulnerabilities related to software and dependencies. GitLab is a CVE Numbering Authority (CNA). Follow this page for generating CVE ID Requests.
GitLab supports the sending of email notifications to users from the GitLab application instance. DHS BOD 18-01 guidance indicates that Domain-based Message Authentication, Reporting & Conformance (DMARC) must be configured for outgoing messages as spam protection. GitLab provides configuration guidance for SMTP across a wide range of email providers, which may be used to help meet this requirement.
Other Services and Concepts
Runners
Runners are required for a wide variety of tasks and tools in any GitLab deployment. To maintain data boundary requirements, customers may need to deploy self-managed runners in their authorization boundary. GitLab provides detailed information on configuring runners, which includes concepts such as:
-
Maximum job timeouts
-
Protecting Sensitive Information
-
Configuring Long Polling
-
Authentication Token Security and Token Rotation
-
Preventing Revealing Sensitive Information
-
Runner Variables
Leveraging APIs
GitLab provides a robust set of APIs to support the application, including REST and GraphQL APIs. Securing APIs starts with the proper configuration of authentication for users and jobs calling the API endpoints. GitLab recommends configuring access tokens (personal access tokens not supported by FIPS) and OAuth 2.0 tokens to control access.
Extensions
Extensions may meet NIST 800-53 requirements depending on which integrations are established. Editor and IDE extensions, for example, may be permissible whereas integrations with third parties may violate authorization boundary requirements. It is the customer's responsibility to validate all extensions to understand where data is being sent outside of the customer's authorization boundary.
Additional Resources
GitLab provides a hardening guide for self-managed customers that covers topics such as:
GitLab CIS Benchmark Guide - GitLab has published a CIS Benchmark to guide hardening decisions in the application. This may be used in concert with this guide to harden the environment in accordance with NIST 800-53 controls. Not all suggestions in the CIS Benchmark directly align with NIST 800-53 controls, but serve as best-practices for maintaining a GitLab instance.