docs(todo): migrate VPS2 IP to domain across registry references

98 operational files still carry 146.59.87.168 — the domain was only adopted
for the git remote, not for container registry references. Bulk is app
manifests' image: lines, plus .gitmodules, both CI workflows, the signed
catalog.json, and two Android companion files with compiled constants. The 117
hits in .planning/ are historical records and stay.

Not a find-and-replace: the domain serves Gitea over HTTPS:443 while images are
pulled from :3000 over plain HTTP, and podman treats host:3000 and domain as
different registries — so every node re-pulls under the new name and any node
that can't resolve or trust the new host fails to pull. It also invalidates the
signed catalog (needs a re-sign ceremony) and the APK ships compiled constants.

Rollout order: registry serving on the domain → manifests → catalog re-sign →
APK rebuild. Steps 2-4 are actively breaking until step 1 holds.

Analysis from the concurrent agent's session before it ended; recorded so it is
not lost.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

.planning/todos/pending/2026-08-02-migrate-vps2-ip-to-domain-in-registry-references.md

@@ -0,0 +1,59 @@
+---
+created: 2026-08-02T12:30:00.000Z
+title: Migrate VPS2 IP (146.59.87.168) to its domain across registry references
+area: infra
+severity: major
+files:
+  - apps/*/manifest.yml (the bulk — every `image:` line)
+  - .gitmodules
+  - .github/workflows/*.yml (both)
+  - app-catalog/catalog.json (signed — changing it forces a re-sign)
+  - Android/**/FipsPreferences.kt, Android/**/PartyScreen.kt (compiled constants)
+---
+
+## Problem
+
+**98 operational files still carry the bare IP `146.59.87.168`.** The domain
+(`source.archipelago-foundation.org`) was only ever adopted for the *git remote* — the IP is
+still baked into every **container registry reference**, which is a different thing entirely:
+
+```
+image: 146.59.87.168:3000/lfg2025/gatewayd:v0.10.0
+```
+
+Plus `.gitmodules`, both CI workflow files, `app-catalog/catalog.json`, and the Android
+companion (`FipsPreferences.kt`, `PartyScreen.kt`). A further **117 hits live in `.planning/`
+docs — those are historical records and must stay as they are.**
+
+An IP in every manifest is exactly the kind of thing that bites when the VPS moves. It is also
+conspicuous in a public repo, so the open-source-readiness work will want it done
+(see `docs/OPEN-SOURCE-READINESS-PLAN.md`).
+
+## Why this is not a find-and-replace
+
+1. **The registry has to actually answer on the domain.** `source.archipelago-foundation.org`
+   currently serves Gitea over **HTTPS on 443**, while images are pulled from **:3000 over plain
+   HTTP**. Podman treats `host:3000` and `domain` as *different registries*, so changing the
+   string means every node re-pulls every image under the new name — and any node that cannot
+   resolve or trust the new host fails to pull at all.
+2. **It invalidates the signed catalog.** `app-catalog/catalog.json` carries image references,
+   so changing them breaks the signature and forces a regenerate-and-re-sign — which needs the
+   release mnemonic, same as a release.
+3. **The Android companion ships compiled constants**, so it needs its own APK rebuild to follow.
+
+## Solution
+
+Its own plan with a real rollout order — **not** something to slip into a release:
+
+1. Registry serving on the domain (TLS, and a decision on whether images move to 443 or the
+   domain also exposes :3000)
+2. Manifests
+3. Catalog regenerate + re-sign (mnemonic ceremony)
+4. APK rebuild
+
+Sequencing matters because steps 2–4 are useless — and actively breaking — until step 1 holds.
+
+## Notes
+
+Analysis produced by the concurrent companion/release agent on 2026-08-02 before its session
+ended; recorded here so it is not lost. It had not touched any of the work.