mi/werkator
1
0
Fork 0
Code Issues Pull Requests 1 Actions Packages Projects Releases Wiki Activity

Merge pull request #10 from mhoennig/22-multi-repo

Browse Source 
ADR 0009 and step 22 plan: one Werkator instance serves a set of repositories

# Conflicts:
#	docs/plan/22-multi-repo.md
#	docs/prs/2026-09-01-PR#6-werkdock-bootstrap.md
This commit is contained in:
mhoennig
2026-09-02 07:22:59 +02:00
parent a1de3c861f 6a5e843b32
commit 713edf77d6
4 changed files with 191 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/adrs/0009-2026-09-01.multi-repo-instance.md
+60
View File
@@ -0,0 +1,60 @@
# One Werkator Instance Serves a Set of Repositories
**Status:**
- proposed: 2026-09-01
- accepted: 2026-09-01
- rejected: -
- superseded: -
**Decision [accepted]:** The founding tenet "one instance per repository" is revised to "one instance per repository *set*": a Werkator instance aggregates self-contained repositories listed in an instance registry, sharing one service, one port, one UI, one watcher schedule, and one global executor cap.
Implementation is planned as `docs/plan/22-multi-repo.md`; this ADR records the decision and its shape, not the code.
## Context and Problem Statement
"One instance per repository" (docs/Werkator-Konzept.md) does not scale even to two repositories on a Hostsharing Managed Webspace: building Werkbaum next to Werkator on mih34 would need a second service, a second assigned port, a second tunnel or domain, a second UI and metrics page.
Every repository multiplies operations while the instance-level resources could be shared.
### Technical Background
Everything repository-specific already lives inside the repository or is keyed by it: machine config with secrets, build results, auto-build slots, worktrees, and buildenvs under `.git/werkator/`; the artifact store under a per-repo key.
An instance can therefore aggregate repositories without absorbing their state — adding or removing a repository is a registry entry, never a data migration, and single-repo mode stays the degenerate case (a registry of one, implicitly the current working directory).
## Considered Options
* One instance per repository set (registry + aggregation) — chosen
* A federation dashboard proxying several single-repo instances
* Status quo: one instance, one repository
### One Instance per Repository Set
Good:
- One service, port, UI, tunnel/domain, metrics page for any number of repositories.
- Repositories stay self-contained; per-repo secrets, history, and pinning semantics are untouched.
- Global concurrency control across all repositories.
Bad:
- The repo dimension must be threaded through executor pools, watcher cycle, routes, and UI — the largest refactor since the rewrite (mitigated by a behavior-preserving `RepoContext` session first).
- Instance-level and repo-level configuration must be split cleanly (see below).
### Federation Dashboard over Single-Repo Instances
Rejected: less invasive, but it keeps N services, N ports, and N tunnels — it solves only the UI aggregation, not the operations burden that motivated the change.
### Status Quo
Rejected: on a webspace, ports and domains are the scarce, manually assigned resource; per-repository services do not scale there.
## Decision Outcome
One instance per repository set, with the configuration split decided 2026-09-01:
- **Instance config** lives in `~/.werkator.yml` in the home directory of the user running the instance — one instance per OS user, matching the platform model. It carries `server.*` (port, domain/public base URL, nginx), the repository registry, the control token, the global `executor.maxConcurrent`, and the watcher schedule. The file name stays `.werkator.yml` in all three locations; the location carries the meaning.
- **Repo defaults** may live in the home file, but only in an explicit `defaults:` block, merged *below* every repository's own layers (home defaults → committed project config → repo machine config → branch layer); pinning semantics are unchanged. Accepted cost: secrets may then live in two places.
- **Repo config** stays in each repository: the committed `.werkator.yml` and the machine config in its `.git/werkator/`.
- Instance keys found in a repo's machine config are ignored with a warning naming both files once a home config exists — never merged silently.
- When a home config with a registry exists, `werkator server` serves the registry regardless of the current directory; without one it serves the current directory exactly as before.
- Repository names (routes, UI) default to the directory basename, are overridable per registry entry, and duplicates abort the start loudly.
Consequences: `docs/Werkator-Konzept.md` and AGENTS.md change their wording when the implementation lands (plan step 22 sessions B–D); until then this ADR documents the target and the existing behavior remains accurate.
docs/plan/22-multi-repo.md
+86
View File
@@ -0,0 +1,86 @@
# Step 22: One Werkator Instance, Many Repositories
Prerequisites: none in code; step 21's Werkdock work is independent.
Read `README.md` first.
This step is a roadmap in sessions (A–E), like step 21; each session is sized for one focused Claude Code session.
## The Problem
"One instance per repository" is a founding tenet (`docs/Werkator-Konzept.md`, AGENTS.md) — and on a Managed Webspace it does not scale even to two repositories.
Building Werkbaum next to Werkator on mih34 today means: a second pac user or a second service, a second assigned port, a second tunnel or domain, a second UI, a second metrics page.
Every repository added multiplies operations, while the instance-level resources (port, UI, watcher schedule, executor slots, metrics) could be shared.
The goal: one Werkator instance serves a *set* of repositories — one service, one port, one UI — while each repository keeps its own configuration, secrets, history, and artifacts.
## The Guiding Idea: the Repository Stays Self-Contained
Everything repository-specific already lives *inside* the repository: the machine config with secrets in `.git/werkator/`, build results, auto-build slots, worktrees, buildenvs — and the artifact store is already keyed per repo path.
The multi-repo instance therefore does not absorb repository state; it becomes an *aggregator* over self-contained repositories.
Consequences:
- Adding or removing a repository is editing a registry entry, never a data migration.
- A repository can move between instances (or back to its own) without losing anything.
- Single-repo mode stays the degenerate case: a registry of one, implicitly the current working directory — existing installations keep working without any config change.
## Key Ownership Splits
Today all config comes from the repo's own layers; multi-repo splits ownership:
- **Instance-level** (decided 2026-09-01: a `.werkator.yml` in the *home directory* of the user running the instance — the name stays `.werkator.yml` in all three locations, the location carries the meaning): `server.*` (port, bind address, public base URL / domain, nginx), the repository registry, the control token (one UI, one token), `executor.maxConcurrent` as the *global* cap, watcher interval, metrics.
- **Repo defaults** (decided 2026-09-01): the home file MAY carry defaults for repo-level keys (e.g. one `git.account`/`git.token` for all repos of the same forge), in an explicit `defaults:` block so instance keys and repo defaults never blur syntactically.
The block merges BELOW every repo's own layers: home `defaults` → committed project config → repo machine config → branch layer (pinning semantics unchanged — home and repo machine config are both host-side layers, the branch layer still cannot reach pinned keys).
Accepted cost: secrets may then live in two places; a repo without its own secrets is no longer self-contained on its own.
- **Repo-level** (unchanged, from the repo's own layers — machine config in its `.git/werkator/`, committed project config, branch layer): `gitea.*` (each repo has its own owner/repo/token/statusContext), `git.*` credentials, `builds`, retention, per-repo watcher options (e.g. `pullRequestGate`), sandbox policy and its pinning.
- **Both**: a per-repo concurrency cap below the global one may come later; not in the first cut.
The pinning model is untouched: pinned keys still come from each repo's machine config, and the branch layer still cannot reach them.
## The Sessions
### A — Decision and schema (ADR 0009)
- ~~Write ADR 0009~~ — done 2026-09-01: `docs/adrs/0009-2026-09-01.multi-repo-instance.md` revises the one-instance-per-repository tenet to one-instance-per-*set* and records the aggregator idea, the key ownership split, the four 2026-09-01 decisions, and the rejected federation-dashboard alternative.
- Define the instance config: `~/.werkator.yml` (decided 2026-09-01) — the repository registry plus the instance-level keys above; one instance per OS user, which matches the platform model (pac users on a webspace, service users elsewhere).
`werkator server` without a home config serves the current directory exactly as today.
- Decide the transition for instance keys that today sit in a repo's machine config (mih34's carries `server.*`): once a home config exists, repo-level instance keys are ignored with a warning naming both files — never merged silently.
- Repo identity for display and routes (decided 2026-09-01): a short unique name per registry entry, defaulting to the repository's directory basename, overridable in the entry; duplicate resulting names abort the start loudly. Used as the route segment (`/repos/<name>/…`) and UI grouping key.
- Precedence (decided 2026-09-01): when a home config with a registry exists, `werkator server` serves the registry regardless of the current directory — one user, one instance, deterministic; without a home config it serves the current directory exactly as today.
- Update `docs/Werkator-Konzept.md` and the AGENTS.md architecture wording ("one instance per repository set") when the implementation lands (sessions B–D) — until then the existing behavior description remains accurate; the AGENTS.md decision list carries ADR 0009 already.
### B — RepoContext refactor, behavior unchanged
- Introduce a `RepoContext` (working dir, config loading, git access, result repository, artifact store key, watcher state) and thread it through executor, watcher, and server code paths that today implicitly use the single `workingDir`.
- The executor becomes instance-global with repo-scoped pools: serialization per (repo, branch), the global `maxConcurrent` across repos; `BuildResult` needs no schema change — results stay in each repo's own JSON file, the repo dimension exists only in memory and in routes.
- Single-repo behavior, routes, and UI stay byte-identical; the full test suite is the acceptance gate.
### C — The registry and N repositories
- Load the registry, build one `RepoContext` per entry; fail the start loudly on duplicate names or unreadable repos (config-version violations abort only that repo's registration, like branch-config violations fail only that branch).
- Watcher multiplexing: one poll cycle iterates the contexts (fetch, enqueue, prune per repo) with per-repo error isolation — one unreachable origin must not starve the others; `WatcherState` gains the repo dimension for the health banner.
- Startup recovery per repo; auto-build slots stay in each repo's `.git/werkator/`.
- CLI commands gain an optional repo selector and default to the current working directory, so `werkator status` inside a repo behaves as today.
### D — Server, API, and UI scoping
- Routes gain the repo segment (`/api/repos/<name>/builds/…`, `/repos/<name>/builds/<key>`); with exactly one registered repo the today-routes keep working (redirect or alias) so bookmarks and posted Gitea links survive.
- Latest/branches/history views group by repo or gain a repo column; one instance-wide metrics page; one control token.
- Gitea status links use the repo-scoped URLs.
### E — Rollout on mih34: Werkbaum joins
- Registry with the Werkator and Werkbaum repositories under the existing user, one service, one port, the existing tunnel.
- Write Werkbaum's `.werkator.yml`: Gradle backend build and npm frontend build in the shared trimmed image (Node is already in it).
- Record the deployment; retire the second-instance/second-user idea from the notes.
## Open Questions
- Fairness across repos when the global concurrency cap is contended (round-robin per repo vs. FIFO) — decide in session C with the real queue behavior at hand.
- Whether buildenv rootfs trees should be shared across repos (today each repo unpacks its own under `.git/werkator/buildenv/`) — the natural answer is Werkdock's image store (step 21 session C), not instance-level state; until then duplicate unpacked rootfs trees are the accepted cost.
- Whether `artifactKey` needs a repo prefix or stays globally unique by construction (random suffix) — decide in session B when the routes are designed.
## Acceptance Criteria
- Session A: ADR 0009 written (done 2026-09-01); the registry and key ownership land in `docs/configuration.md` together with the implementing sessions, since that reference describes implemented configuration only.
- Session B: full suite green with `RepoContext` threaded through; no route or behavior change observable.
- Session C: an instance with two registered repos builds pushes in both, with per-repo error isolation proven by a test (one broken origin, the other keeps building).
- Session D: both repos browsable in one UI; single-repo installations keep their existing URLs.
- Session E: mih34 builds Werkator and Werkbaum from one service; `docs/deployment.md` describes the registry setup.
docs/prs/2026-09-02-PR#10-step-22-multi-repo-decision.md
+44
View File
@@ -0,0 +1,44 @@
> **WARNING:** This document describes only the change applied in this PR.
> It may already be outdated once the next PR is merged.
> Historic PR-documentation is not maintained along with new PRs — treat it as a snapshot, not as current documentation.
## The Problem
"One instance per repository" is a founding tenet (`docs/Werkator-Konzept.md`, AGENTS.md), and it does not scale even to two repositories on a Hostsharing Managed Webspace: building Werkbaum next to Werkator on mih34 would need a second service, a second assigned port, a second tunnel or domain, a second UI, a second metrics page.
Every repository added multiplies operations, while the instance-level resources (port, UI, watcher schedule, executor slots, metrics) could be shared.
## Non-Goals
- Implementing the refactor — this PR is the decision and the roadmap only; the sessions it lays out (B–E: `RepoContext` refactor, registry/watcher multiplexing, routes/UI scoping, mih34 rollout with Werkbaum) are future PRs.
- Changing `docs/Werkator-Konzept.md`'s or AGENTS.md's architecture wording — that happens when the implementation lands, not with the decision.
## The Solution
ADR 0009 revises the tenet to "one instance per repository *set*": a Werkator instance aggregates self-contained repositories listed in an instance registry, sharing one service, one port, one UI, one watcher schedule, and one global executor cap, while each repository keeps its own configuration, secrets, history, and artifacts — nothing repository-specific moves out of the repository, so adding or removing one is a registry entry, never a data migration.
Config ownership splits three ways: **instance-level** config (`server.*`, the registry, the control token, the global `executor.maxConcurrent`, watcher schedule) lives in `~/.werkator.yml` in the home directory of the user running the instance — one instance per OS user, matching the platform's pac-user model; the file name stays `.werkator.yml` everywhere, only the location carries the meaning.
**Repo defaults** may live in the same home file, but only in an explicit `defaults:` block merged *below* every repository's own layers (home defaults → committed project config → repo machine config → branch layer) — pinning semantics are unchanged, at the accepted cost that secrets may then live in two places.
**Repo-level** config (`gitea.*`, `git.*` credentials, `builds`, retention, sandbox policy) stays exactly where it is today, in each repository's own layers.
Four follow-on decisions were folded in during review: repo-level instance keys found once a home config exists are ignored with a warning naming both files, never merged silently; when a home config with a registry exists it is served regardless of the current directory (registry wins over cwd); repository names for routes/UI default to the directory basename, are overridable per entry, and duplicates abort the start loudly; the `.werkator.yml` name is kept in all three locations rather than inventing a separate instance-config filename.
A federation dashboard proxying several single-repo instances was considered and rejected: it solves only UI aggregation, not the per-repository operations burden (ports, tunnels, services) that motivates the change.
## Open Questions
- Fairness across repos when the global concurrency cap is contended (round-robin vs. FIFO) — deferred to session C, decided with the real queue behavior at hand.
- Whether buildenv rootfs trees should be shared across repos, or whether that is better solved by Werkdock's image store — deferred; duplicate unpacked rootfs trees are the accepted interim cost.
- Whether `artifactKey` needs a repo prefix or stays globally unique by construction — deferred to session B, when routes are designed.
## Additional Changes
- `docs/plan/22-multi-repo.md`: the full five-session roadmap (A–E).
- AGENTS.md: decision list gained ADR 0009.
## Prerequisite PRs
- None in code; branches from `werkdock-extraction` (PR #6) but is otherwise independent of the Werkdock/webspace work in PR #7/#8/#9.
## Follow-up PRs
- Session B: `RepoContext` refactor, behavior-preserving.
- Session C: the registry and N repositories, watcher multiplexing.
- Session D: server/API/UI repo scoping.
- Session E: rollout on mih34 with Werkbaum joining the instance.