ngit-grasp/closed/9f23-archive-services-setting.md

Archive Services Setting

ID: 9f23

Problem

Operators need a way to backup/archive one or more GRASP servers by domain. Currently, there's no configuration option to selectively accept announcement events based on the services they advertise.

We need an archive-services setting that:

• Accepts a list of domain names (GRASP server domains to archive)
• Filters announcement event acceptance based on whether the event lists any of the configured services
• Enables relay operators to act as archival mirrors for specific GRASP servers

Plan

• Phase 1: Add archive_grasp_services configuration option
  • Update src/config.rs with new field
  • Update docs/reference/configuration.md
  • Update nix/module.nix
  • Update .env.example
• Phase 2: Implement announcement event filtering logic
  • Extract service domains from announcement events
  • Compare against configured archive-grasp-services list
  • Accept/reject based on match
• Phase 3: Add tests for archive-grasp-services functionality
  • Unit tests for configuration parsing
  • Integration tests for event filtering

Progress

2026-01-21 [Session Start]

• Started: Created issue for archive-services setting feature
• Started work: Created worktree 9f23-archive-services-setting
• Completed Phase 1: Added archive-grasp-services configuration option
  • Implemented in all four sources (config.rs, docs, nix module, .env.example)
  • Added validation for mutual exclusivity with archive-all and archive-whitelist
  • Defaults read-only to true
  • Added 9 comprehensive tests (all 61 config tests pass)
• Completed Phase 2: Implemented filtering logic
  • Added clone_url_domains() to extract domains from announcement clone URLs
  • Added matches_grasp_services() for case-insensitive domain matching
  • Integrated filtering into validate_announcement() flow
  • Added 8 tests for domain extraction and matching (all 390 lib tests pass)
• Decision: Only bare domains accepted (no ws:// or wss:// prefixes)
  • Clone URLs use https:// scheme, domain extraction strips that prefix
  • Exact case-insensitive matching ensures correct behavior
  • Documentation updated to clarify bare domain requirement
• Completed Phase 3: Added integration tests
  • Created tests/archive_grasp_services.rs with 4 test scenarios
  • Tests matching/non-matching domains, multiple services, case-insensitivity
  • All 39 integration tests pass (525 total tests)
• Simplified implementation via rebase:
  • Rebased onto master with clean commit history
  • Removed complex matches_grasp_services() function
  • Reused existing lists_service() infrastructure
  • Added has_clone_url_with_case() helper for case-insensitive matching
  • Validation logic now uses existing methods: checks clone URLs only (archiving from other services)
  • Removed 88 lines of unnecessary code, added 42 lines of simple helpers
  • Net improvement: 46 lines less code, much cleaner
• Final commits: 25351fd (test helper fix), f911d00 (complete feature)
• All 39 integration tests pass

2026-01-21 [Session End]

• Completed: All phases finished, tests passing, ready to merge
• Merging back to main and cleaning up worktree

Notes

• This enables relay operators to selectively mirror specific GRASP servers
• Announcement events contain service information that can be matched against the configured list
• Must maintain consistency across all four configuration sources (critical per AGENTS.md)