lhumina_code/home
21
0
Fork 0
Code Issues 10 Pull requests Projects Releases Packages Wiki Activity Actions

hero_services: Docker deployment broken after service config restructuring #49

New issue
Closed
opened 2026-03-19 19:19:10 +00:00 by mik-tf · 3 comments
mik-tf commented 2026-03-19 19:19:10 +00:00
Owner
Copy link

Problem

The recent merge of development_kristof into development restructured service configs but left the Docker deployment pipeline broken. This is part of the larger hero_proc transition (#50).

What's broken

  1. Profiles not copied to Docker image — profiles/ directory exists in repo but docker/build-local.sh never copies it to dist/. The entrypoint runs hero_services_server run --profile user but the binary can't find profiles.

  2. build-local.sh validation — still expects services/user/ (old layout). Configs are now flat in services/.

  3. Binary name mismatch — service TOMLs reference new names (_openrpc/_http) but most repos still produce old names (_server/_ui). This is expected during the transition per #50.

Specific files

  • docker/build-local.sh — needs to copy profiles/ to dist/, fix validation for flat layout
  • docker/entrypoint.sh — needs binary symlinks (_server → _openrpc, _ui → _http) as a bridge
  • Dockerfile.pack — needs COPY profiles/ line

Context

Per #50, the architecture is transitioning to:

  • hero_proc replacing zinit
  • Services self-registering with --start flag
  • OpenRPC-first naming (_openrpc/_http)
  • No more profile subdirectories

This fix bridges the current state until the full transition completes.

Impact

  • Local Docker builds fail (no services start)
  • herodev will break on next deploy
  • Any fresh container from development branch is non-functional
## Problem The recent merge of `development_kristof` into `development` restructured service configs but left the Docker deployment pipeline broken. This is part of the larger hero_proc transition (#50). ### What's broken 1. **Profiles not copied to Docker image** — `profiles/` directory exists in repo but `docker/build-local.sh` never copies it to `dist/`. The entrypoint runs `hero_services_server run --profile user` but the binary can't find profiles. 2. **build-local.sh validation** — still expects `services/user/` (old layout). Configs are now flat in `services/`. 3. **Binary name mismatch** — service TOMLs reference new names (`_openrpc`/`_http`) but most repos still produce old names (`_server`/`_ui`). This is expected during the transition per #50. ### Specific files - `docker/build-local.sh` — needs to copy `profiles/` to `dist/`, fix validation for flat layout - `docker/entrypoint.sh` — needs binary symlinks (`_server` → `_openrpc`, `_ui` → `_http`) as a bridge - `Dockerfile.pack` — needs `COPY profiles/` line ### Context Per #50, the architecture is transitioning to: - hero_proc replacing zinit - Services self-registering with `--start` flag - OpenRPC-first naming (_openrpc/_http) - No more profile subdirectories This fix bridges the current state until the full transition completes. ### Impact - Local Docker builds fail (no services start) - herodev will break on next deploy - Any fresh container from development branch is non-functional
mik-tf commented 2026-03-19 20:37:15 +00:00
Author
Owner
Copy link

Progress: compile fixes pushed to development

All 18 services now compile (0 failures). Fixes pushed across 5 repos:

Repo Commit Fix
hero_services 610e18c Revert TOML binary names (_openrpc→_server, _http→_ui), fix build-local.sh flat layout + profiles copy, add profiles to Dockerfile.pack, add entrypoint socat bridge + service recovery
hero_redis 873baa3 Remove 927 lines of dead dispatch code from botched merge, fix RedisServer::new signature
hero_foundry 80c8bcc Resolve committed merge conflict markers in hero_foundry_admin
hero_osis 4f07f72 Restore Video/VideoPlaylist support (611 lines) lost during HeroRpcServer migration merge
hero_indexer 20a844c Update Cargo.lock for hero_proc_sdk (stale commit from force-push)

What works

  • All 18 services compile and build via SKIP_WASM=1 make dist
  • Docker image builds successfully
  • Services run correctly when started individually inside the container
  • Video feature fully restored (schema, SDK, handlers, server glue all intact)

What remains: hero_proc migration

The Docker deployment still uses zinit as the process supervisor, which causes startup race conditions (services exhaust retry limits before dependencies stabilize). Per #50, hero_proc replaces zinit entirely. Next steps:

  1. Clone hero_proc into lhumina_code workspace
  2. Replace zinit_sdk with hero_proc_sdk in hero_services_server
  3. Replace zinit_server with hero_proc_server in Docker image
  4. Update entrypoint.sh to use hero_proc orchestration
  5. Update build-local.sh to build hero_proc instead of zinit

This is tracked under #50. Keeping this issue open until Docker deployment is fully functional with hero_proc.

## Progress: compile fixes pushed to development All 18 services now compile (0 failures). Fixes pushed across 5 repos: | Repo | Commit | Fix | |---|---|---| | hero_services | 610e18c | Revert TOML binary names (_openrpc→_server, _http→_ui), fix build-local.sh flat layout + profiles copy, add profiles to Dockerfile.pack, add entrypoint socat bridge + service recovery | | hero_redis | 873baa3 | Remove 927 lines of dead dispatch code from botched merge, fix RedisServer::new signature | | hero_foundry | 80c8bcc | Resolve committed merge conflict markers in hero_foundry_admin | | hero_osis | 4f07f72 | Restore Video/VideoPlaylist support (611 lines) lost during HeroRpcServer migration merge | | hero_indexer | 20a844c | Update Cargo.lock for hero_proc_sdk (stale commit from force-push) | ### What works - All 18 services compile and build via `SKIP_WASM=1 make dist` - Docker image builds successfully - Services run correctly when started individually inside the container - Video feature fully restored (schema, SDK, handlers, server glue all intact) ### What remains: hero_proc migration The Docker deployment still uses **zinit** as the process supervisor, which causes startup race conditions (services exhaust retry limits before dependencies stabilize). Per #50, **hero_proc replaces zinit entirely**. Next steps: 1. Clone hero_proc into lhumina_code workspace 2. Replace zinit_sdk with hero_proc_sdk in hero_services_server 3. Replace zinit_server with hero_proc_server in Docker image 4. Update entrypoint.sh to use hero_proc orchestration 5. Update build-local.sh to build hero_proc instead of zinit This is tracked under #50. Keeping this issue open until Docker deployment is fully functional with hero_proc.
mik-tf commented 2026-03-19 21:46:33 +00:00
Author
Owner
Copy link

hero_proc migration complete — runtime working

Pushed to development: bd14ea1 — feat: replace zinit with hero_proc as process manager (#50)

What was done

  • Replaced zinit_sdk with hero_proc_sdk across hero_services_server
  • Docker entrypoint starts hero_proc_server instead of zinit_server
  • build-local.sh builds hero_proc workspace instead of zinit
  • bootstrap.sh clones hero_proc from lhumina_code (no more geomind_code/zinit)
  • Fixed profile registration ordering for hero_proc (pass-based: install → run → health → test)
  • Fixed service TOML exec lines (binary names, serve subcommand, removed hero_fossil)
  • Added hero_proc-based recovery pass in entrypoint.sh

Test results

  • 21/21 services running in Docker
  • Hero OS login page serves at localhost:8080
  • No crash loops

Remaining

5 services still use old CLI pattern (no serve subcommand) — tracked in #52.

Signed-off-by: mik-tf

**hero_proc migration complete — runtime working** Pushed to development: `bd14ea1` — feat: replace zinit with hero_proc as process manager (#50) ### What was done - Replaced zinit_sdk with hero_proc_sdk across hero_services_server - Docker entrypoint starts hero_proc_server instead of zinit_server - build-local.sh builds hero_proc workspace instead of zinit - bootstrap.sh clones hero_proc from lhumina_code (no more geomind_code/zinit) - Fixed profile registration ordering for hero_proc (pass-based: install → run → health → test) - Fixed service TOML exec lines (binary names, serve subcommand, removed hero_fossil) - Added hero_proc-based recovery pass in entrypoint.sh ### Test results - 21/21 services running in Docker - Hero OS login page serves at localhost:8080 - No crash loops ### Remaining 5 services still use old CLI pattern (no serve subcommand) — tracked in #52. Signed-off-by: mik-tf
mik-tf commented 2026-03-20 00:50:00 +00:00
Author
Owner
Copy link

Docker deployment working — 20/21 services running, all smoke tests pass. Fixed in P0 and verified through P1-P2 work.

Signed-off-by: mik-tf

Docker deployment working — 20/21 services running, all smoke tests pass. Fixed in P0 and verified through P1-P2 work. Signed-off-by: mik-tf
Sign in to join this conversation.
No Branch/Tag specified
Branches Tags
development
main
No results found.
Labels
Clear labels
No items
No labels
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
lhumina_code/home#49
No description provided.