Replace sensitive zone_id variable with hardcoded value and update
DNS record configuration to use cloudflare_dns_record resource type.
Disable proxying and set explicit TTL for better control over DNS
propagation.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Add documentation about required OCI and AWS credentials for the
trantor configuration, clarifying that ~/.oci/config can be used
as an alternative to environment variables.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Configure global DNS nameservers for the Tailscale tailnet, setting
trantor as the primary DNS server with Cloudflare as fallback. This
enables custom DNS resolution across the entire tailnet.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Added Kanidm identity provider integration with Nextcloud:
- Enabled Kanidm client in kanidm.nix for CLI access
- Added user_oidc app to Nextcloud for OpenID Connect authentication
- Configured allow_local_remote_servers to permit Nextcloud to reach
Kanidm at auth.baduhai.dev (resolves to local IP 192.168.15.142)
OAuth2 client configuration (done via kanidm CLI):
- Client ID: nextcloud
- Scopes: openid, email, profile mapped to idm_all_accounts group
- Redirect URI: https://cloud.baduhai.dev/apps/user_oidc/code
- User mapping: name claim maps to Nextcloud username
This allows users to authenticate to Nextcloud using their Kanidm
credentials, with existing Nextcloud accounts linked via username.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Changed certificate generation from HTTP-01 to DNS-01 challenge to support
services behind Tailscale/CGNAT IPs. HTTP-01 challenges fail because Let's
Encrypt cannot reach private Tailscale IPs (100.x.x.x) that Cloudflare DNS
points to.
Changes:
- Pre-configure certificates in security.acme.certs using DNS-01 via Cloudflare
- Auto-generate certificate configs from shared/services.nix
- Alexandria: filters services with host == "alexandria"
- Trantor: filters services with host == "trantor"
- Updated mkNginxVHosts to use useACMEHost instead of enableACME
- Each domain gets its own certificate configured with DNS-01 challenge
This ensures all services get valid Let's Encrypt certificates even when
accessible only through Tailscale or private networks.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Added Kanidm server configuration to serve as central identity provider for
all services. Configuration includes:
- Server on auth.baduhai.dev with HTTPS
- LDAP support on port 636 for legacy integrations
- Nginx reverse proxy with SSL termination
- Added to shared services for DNS resolution
Kanidm will provide OAuth2/OIDC authentication for Nextcloud, Vaultwarden,
Forgejo, and other services.
Updated mkNginxVHosts to use per-domain certificates (enableACME) instead
of shared wildcard certificates (useACMEHost). Each service now requests
its own certificate, avoiding conflicts between hosts and following the
principle of least privilege.
Removed wildcard certificate configuration from both alexandria and trantor.
Each host now only obtains certificates for domains it actually serves:
- Alexandria: pass.baduhai.dev, cloud.baduhai.dev, jellyfin.baduhai.dev
- Trantor: git.baduhai.dev
Migrated forgejo service and configuration to trantor. Added nginx reverse
proxy support on trantor with ACME configuration for SSL certificates.
Fixed vaultwarden domain in shared services from vault.baduhai.dev to
pass.baduhai.dev to match actual nginx configuration.
Alexandria's unbound now only serves LAN clients (192.168.0.0/16) and
returns LAN IPs for service domains.
Created new unbound instance on trantor to serve Tailscale clients
(100.64.0.0/10) and return tailscale IPs for service domains.
Both configurations pull service records from shared/services.nix.
Removed the split-dns.nix module and all service-specific splitDNS.entries
configurations. Service DNS records are now sourced from the centralized
shared/services.nix file instead of being declared individually in each
service configuration.
Created centralized service definitions in shared/services.nix to store
service metadata (domains, IPs, ports) that need to be accessible across
multiple hosts. This replaces the per-service split DNS module approach
with a single source of truth.
Services are now exported through utils.nix for easy access in host configs.
