Repository navigation
ADR: Split platform configuration into public and private repositories #49
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
base: main
Are you sure you want to change the base?
Changes from all commits
File filter
Filter by extension
Conversations
Jump to
Diff view
Diff view
There are no files selected for viewing
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,46 @@ | ||
| --- | ||
| title: Split platform configuration into public and private repositories | ||
| standards: | ||
| - nist-ac | ||
| - nist-cm | ||
| --- | ||
|
|
||
| ## Context | ||
|
|
||
| The platform is configured from code on GitHub: manifests and infrastructure-as-code for the on-premises OpenShift cluster at the MGHPCC and for the foundational services (Keycloak, LDAP) on AWS EKS. We want that configuration to be open and reviewable, in keeping with an open cloud. | ||
|
|
||
| Some of the values it needs are sensitive, such as IP addresses and network topology. They are not CUI, but publishing them helps an attacker, and NIST SP 800-171 3.1.22 requires control over information posted on publicly accessible systems. | ||
|
|
||
| ## Decision | ||
|
|
||
| We will split configuration into two kinds of repositories. | ||
|
|
||
| | | Public repositories | Private repositories | | ||
| |---|---|---| | ||
| | **Holds** | Manifests, Helm charts and Kustomize bases, and infrastructure-as-code modules, with templated placeholders in place of sensitive values. Source code, CI workflows, and documentation. | The values that fill the placeholders: IP addresses, network topology, hostnames, account identifiers. Realm and directory configuration. | | ||
| | **Never holds** | CUI, credentials, or real sensitive values. | CUI or credentials. | | ||
| | **Visible to** | Anyone. | A named team. | | ||
|
|
||
| Credentials belong in neither type. They live in AWS Secrets Manager and reach the cluster through the External Secrets Operator. | ||
|
|
||
| Both kinds of repository follow the same change control: branch protection with no bypass, two approvals from someone other than the author. | ||
|
|
||
| ## Consequences | ||
|
|
||
| ### Benefits | ||
|
|
||
| - Configuration stays open and reviewable without publishing sensitive values. | ||
| - A single review process covers both kinds of repository. | ||
| - The SDLC policy can state clearly what may be public and what may not. | ||
|
|
||
| ### Changes | ||
|
|
||
| - A change that touches a placeholder and its value needs coordinated pull requests in two repositories, and the two can drift apart. | ||
| - The EKS jobs need a read-only credential for the private repositories, which must be rotated at least every 90 days. | ||
| - The private repositories hold sensitive configuration, so access to them must be limited and reviewed. | ||
| - A real value committed to a public repository stays in its history. It must be treated as exposed and replaced. | ||
|
Member
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. I'm not sure I agree with this for the types of information we're talking about here. If we accidentally expose an ip address, we have to..."replace it"? These are not security credentials, and in our current environment addresses are almost exclusively assigned dynamically.
Contributor
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. Maybe we can grandfather in the existing IPs that have been documented? Once the private repo is up, you'd have to go out of your way to expose IPs in the public repo. For the sake of the compliant environment though, I think this rotation is necessary. |
||
|
|
||
| ## See also | ||
|
|
||
| - [NIST SP 800-171 Rev 2: 3.1.22, 3.4.3, 3.4.5](https://csrc.nist.gov/pubs/sp/800/171/r2/upd1/final) | ||
| - [BU Data Classification Standard](https://www.bu.edu/policies/data-classification-standard/) | ||
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This, at least, I think is a solved problem: we use github apps to acquire repository credentials in CI workflows, and those credentials are ephemeral: they existing only for the duration of the workflow.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
In this case we'd need a private key to sign the JWT that we're sending to GitHub, so we'd still need a credential that we rotate every 90 days. https://docs.github.com/en/apps/creating-github-apps/authenticating-with-a-github-app/managing-private-keys-for-github-apps