Add documentation for manual archive removal process

[mprpic]

Document removing archived content upon request.

Requirements:

• Document S3 object removal process
• Create database cleanup procedures
• Document CNA request handling workflow
• Create removal audit trail process (tracked via GitHub issues?)
• Add removal request template and contact information (tracked via GitHub issues?)

Acceptance Criteria:

• Clear documentation for manual content removal
• Database queries for identifying content to remove
• Audit trail for all removal actions
• Contact process for removal requests

[dwelch2344]

Auth will be a big thing here. Either we've got to support cve-services authentication (sharing JWT's private secret is bad news tho) or provide a mechanism for auth of some kind.

I opened CVEProject/cve-services#1530 to discuss something that could be a potential short term solution. And yes, I see the hypocrisy in specifically outlining how it should not be used as auth – but "do as we say, not as we do" kind of applies here since this is being piloted/run by the program.

Edit: 1530 is NOT required for MVP; if anything, makes going public + building controls easier overall.

[dwelch2344]

Database queries for identifying content to remove

We should probably go a step futher. I'd aim to have both the queries and an endpoint that would:

1. On POST /endpoint: Find the resources (and any other downstream effects; at minimum, interval bookkeeping) and return the information, but take no action.
2. On POST /endpoint?confirm=true exact same steps as 1, but actually process the removal.

TLDR: one endpoint, with dry-run default semantics, that will handle the complete removal and cleanup.

Audit trail for all removal actions

Requires auth.

[dwelch2344]

One topic we should also consider: Opt-out mechanics. We don't currently filter by it, but we have the ability to specify "do not archive" at both a Domain and Resource level. That enables the following stories:

• as a Non-Member:
  • I can request my entire domain be scrubbed and no longer be archived.
  • I can request a specific subdomain be scrubbed and no longer be archived.
  • I can request a specific URL/reference be scrubbed.

It would not enable:

• as a Non-Member:
  • I can request a specific domain + resource path/regex be scrubbed and removed.

[dwelch2344]

From AWG today:

In talking about whether we're standing up a "lite" MVP, or production-like (with no guarantees) MVP:

• Can we go to MVP-live? Reqs are:
  • Day-2 Operations (defined + impl'd)
  • Observability of usage (for cost, if nothing else)
  • <TODO: what'd I miss>

Cost-per-retrieval, coupled with A/B testing (or "beta")

Static Costs:

• RDS
• Container running
• S3 Storage
• S3 access cost
• CloudFront/CDN access

Dynamic Costs:

• Scale * Access Cost
• RDS scale
• Container Scale