Skip to content

fix(subdomain-catalog): database.0mpc.com is Adminer UI; fix /etc/hosts for extra-sites#16

Merged
baditaflorin merged 1 commit into
mainfrom
claude/fix-catalog-cert-edge-hosts
May 26, 2026
Merged

fix(subdomain-catalog): database.0mpc.com is Adminer UI; fix /etc/hosts for extra-sites#16
baditaflorin merged 1 commit into
mainfrom
claude/fix-catalog-cert-edge-hosts

Conversation

@baditaflorin
Copy link
Copy Markdown
Owner

Summary

  • config/subdomain-catalog.json: Fixed database.example.com — entry was private-only/private_network (describing the postgres TCP proxy), but the subdomain actually serves the Adminer web UI on runtime-general-lv3:8082, published via nginx edge + Keycloak oauth2-proxy. Corrected to edge-published/edge_oidc.
  • config/certificate-catalog.json: Added database-edge entry — cert audit was failing CRITICAL because the updated catalog entry had no corresponding cert coverage.
  • config/subdomain-exposure-registry.json: Regenerated after catalog fix.
  • docker_runtime role defaults: Extended docker_runtime_public_edge_host_aliases to include public_edge_extra_sites hostnames (changelog, docs, database, minio-console) and the apex domain (public_edge_apex_hostname). Previously these were missing from /etc/hosts on all VMs, causing them to fall through to stale external DNS (51.89.70.16) and failing cert validation.

Verification

  • subdomain_exposure_audit.py --validate → exit 0 (was failing with CRITICAL before)
  • make route-dns-assertion-ledger env=production → ok=67, failed=0 (bypassed audit was needed before)
  • Certificate validation from inside runtime-control-lv3: 47/49 valid (was 41/49). Remaining 2 failures are intentional Tailscale-only services (proxmox UI port 8006, vaultwarden port 443) that use 100.64.0.1 targets by design.

Test plan

  • Run make converge-api-gateway env=production after merge to apply updated /etc/hosts to all VMs
  • Re-run cert validator to confirm 47+ valid

…edge_oidc), not postgres TCP

- subdomain-catalog.json: fix database.example.com entry — exposure was
  "private-only/private_network" but the subdomain is actually the Adminer
  web UI on runtime-general-lv3:8082, published via nginx edge behind
  Keycloak oauth2-proxy (edge-published/edge_oidc)
- certificate-catalog.json: add database-edge entry so the cert audit does
  not flag a CRITICAL missing coverage finding
- subdomain-exposure-registry.json: regenerated after catalog fix
- docker_runtime defaults: extend public_edge_host_aliases to include
  public_edge_extra_sites hostnames (changelog, docs, database, minio-console)
  and the apex domain so /etc/hosts is complete on all VMs — fixes
  cert-validator CONNECTION_FAILED for 5 domains that fell through to
  stale external DNS

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
@baditaflorin baditaflorin merged commit b7e16b5 into main May 26, 2026
0 of 2 checks passed
@baditaflorin baditaflorin deleted the claude/fix-catalog-cert-edge-hosts branch May 26, 2026 15:25
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant