chore: drop shared /run bind mount from docker-compose#613
Merged
Conversation
Each service now writes to its own ephemeral /app/run. Cross-service
file access already goes through Postgres (PlanItem.generated_report_html,
PlanItem.run_zip_snapshot) and worker_plan HTTP endpoints (/runs/{id}/zip,
/runs/{id}/files), so the shared bind mount was redundant and mirrors
awkwardly onto Railway (where it had to be a dedicated volume).
- Remove `./run:/app/run` from worker_plan, worker_plan_database_*,
frontend_multi_user.
- Remove PLANEXE_HOST_RUN_DIR env (only purpose was mapping container
paths back to the host bind mount) from docker-compose, .env examples,
worker_plan/railway.md, AGENTS.md notes, and worker_plan/README.md.
- Simplify build_display_run_dir in worker_plan/app.py now that there is
no host base to map to.
- Update docker-compose.md and docs/docker.md to describe the new
per-container run dir and how to inspect it (docker compose exec or
/runs/{id}/zip).
To inspect run outputs during dev: `docker compose exec worker_plan ls /app/run`
or fetch via worker_plan's HTTP endpoints.
Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>
The note only made sense as a migration marker. Now that no-shared-run is the baseline, it adds noise. Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>
Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>
Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>
Readers were being primed to think they need to mount or configure the run directory. It's an internal, per-container implementation detail. Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>
PLANEXE_RUN_DIR was a relic from when services shared a bind-mounted
/run directory. With each service now writing to its own ephemeral
filesystem, the override serves no purpose — every site just fell
back to the default anyway.
- Inline the default at each call site (worker_plan/app.py,
worker_plan_database/app.py, mcp_cloud/db_setup.py) instead of
os.environ.get("PLANEXE_RUN_DIR", ...).
- frontend_multi_user: drop the dotenv override path, use
planexe_project_root / RUN_DIR directly. (app.config key name
kept — it's an internal Flask dict key, not the env var.)
- Remove PLANEXE_RUN_DIR entry from the DotEnvKeyEnum in both
worker_plan_api and worker_plan_internal.
- Remove ENV PLANEXE_RUN_DIR from mcp_cloud and worker_plan_database
Dockerfiles and from docker-compose.yml (3 services).
- Update worker_plan/README.md, worker_plan/AGENTS.md,
mcp_cloud/README.md, mcp_cloud/worker_fetchers.py docstring to
stop referencing the env var.
Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>
The demo plan is no longer bundled with the app (demos live on the website now). SHOW_DEMO_PLAN was hardcoded to False in both app.py and plan_routes.py, making the entire demo branch in /viewplan unreachable. With that gone, self.planexe_run_dir, app.config['PLANEXE_RUN_DIR'], and the RUN_DIR constant have no remaining readers. - Drop RUN_DIR, SHOW_DEMO_PLAN, self.planexe_run_dir, and the app.config['PLANEXE_RUN_DIR'] assignment from frontend_multi_user/src/app.py. - Drop the SHOW_DEMO_PLAN flag and the demo branch from viewplan in plan_routes.py; drop the now-unused send_file import. Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>
display_run_dir is always str(run_dir) now, so the helper and the intermediate variable were just noise. Response schema still exposes display_run_dir per the API contract. Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>
Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>
The fallback existed for setups where mcp_cloud shared a /run bind mount with a worker. Now that each service writes to its own ephemeral filesystem, mcp_cloud's own /run is always empty and the function always returned None. Purge the code path end to end. - Delete list_files_from_local_run_dir from worker_fetchers.py. - Delete BASE_DIR_RUN constant (its only remaining use). - Drop the call site in handlers.py and the noqa re-exports in app.py. - Drop the mock patch from test_plan_status_tool.py. - Update README artifact resolution order and AGENTS.md fallback docs. Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>
Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>
railway-app
Bot
temporarily deployed
to
PlanExe - multi user / staging
The variable is not read anywhere in the codebase; PLANEXE_POSTGRES_HOST is the real one. Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>
railway-app
Bot
temporarily deployed
to
PlanExe - multi user / staging
Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>
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
1 participant
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
Implements the Railway volume kludge TODO from
docs/plan.md. Each service now writes run outputs to its own ephemeral/app/runinstead of sharing a bind-mounted./rundirectory. Cross-service access already goes through Postgres (PlanItem.generated_report_html,PlanItem.run_zip_snapshot) andworker_planHTTP endpoints (/runs/{id}/zip,/runs/{id}/files), so the shared volume was redundant — and mirrored awkwardly onto Railway, where it had to be a dedicated volume../run:/app/runfromworker_plan,worker_plan_database_*,frontend_multi_userindocker-compose.yml.PLANEXE_HOST_RUN_DIR(only purpose was mapping container paths back to the host bind mount) from compose,.envexamples,worker_plan/railway.md, root/worker AGENTS.md, andworker_plan/README.md.build_display_run_dirinworker_plan/app.py— response fielddisplay_run_diris preserved per the worker API contract, it just always equalsrun_dirnow.docker-compose.mdanddocs/docker.mdwith the new per-container layout and how to inspect outputs (docker compose exec worker_plan ls /app/runor/runs/{id}/zip).Test plan
docker compose configvalidatesdocker compose upstarts the full stack cleanly with norun/on hostfrontend_multi_user, report renders and zip downloads workmcp_cloud(plan_create → plan_status → plan_file_info download URL)🤖 Generated with Claude Code