docs: correct multi-worker scaling claims in deployment docs#445
Merged
Conversation
deployment-modes.md and pipeline-architecture.md described load balancing across multiple worker replicas and hybrid embedded/external worker deployments. Neither is implemented: PipelineManager's job queue is in-process memory, only one worker owns the SQLite-backed document store at a time, and enableWorker/--server-url are mutually exclusive. Only the stateless web/mcp coordinator tier scales horizontally today. Aligns both docs with ARCHITECTURE.md. Fixes #443 Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
There was a problem hiding this comment.
Pull request overview
Updates deployment documentation to remove inaccurate claims about multi-worker/hybrid deployments and align the docs with the current single-worker architecture (in-memory job queue + SQLite-backed store), emphasizing that only the coordinator tier scales horizontally.
Changes:
- Rewrites distributed-mode architecture and scaling guidance to describe horizontal scaling for
web/mcpcoordinators only. - Clarifies that the worker process does not scale horizontally and should be scaled vertically (e.g., via
maxConcurrencyand resources). - Removes/rewrites “hybrid deployment” guidance to reflect that embedded-worker and external-worker modes are treated as mutually exclusive in normal deployments.
Reviewed changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated 4 comments.
| File | Description |
|---|---|
| docs/infrastructure/deployment-modes.md | Reframes distributed-mode architecture and scaling guidance around a stateless coordinator tier + single worker. |
| docs/concepts/pipeline-architecture.md | Updates scaling patterns section to clarify horizontal scaling applies to coordinators, not workers, and removes hybrid deployment claims. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Addresses Copilot review feedback on PR #445: the Distributed Mode intro, Worker Management, and Job Recovery sections still used plural "workers", contradicting the single-worker architecture the PR establishes. Also clarifies that --server-url is a coordinator flag that points at the worker, not a flag the worker itself takes. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
deployment-modes.mdandpipeline-architecture.mdclaimed you can load-balance across multiple worker replicas, or run embedded + external workers together ("hybrid deployment"). Neither is implemented, and it directly contradictedARCHITECTURE.md's "shared worker" language.PipelineManager's job queue is in-process memory (PipelineManager.ts), only one worker process owns the SQLite-backed document store at a time, andenableWorker/--server-urlare mutually exclusive (mcp.ts,web.ts,worker.ts) — so a coordinator always talks to exactly one worker.web/mcpcoordinator tier scales horizontally behind a load balancer; the worker does not (scale it vertically viamaxConcurrencyinstead), in part because SQLite's WAL locking assumes one host's local filesystem, not multiple hosts sharing a file over a network volume.Test plan
npm ci(Node 22) +npx biome checkon the two changed files — both are excluded from Biome's markdown checks, so no formatting driftlint-staged) ran cleanPipelineManager.ts,PipelineFactory.ts,mcp.ts,web.ts,worker.ts, andAppServer.tsFixes #443
🤖 Generated with Claude Code