stroblmeandClaude Opus 5
608d30d884
Five concurrent training nodes, each sizing its thread pool to every core,
left the engine's own event loop unscheduled: the API stopped answering
within 10 s and every client died. The same shape on a GPU deadlocked a run
for 21 minutes at 0% utilisation with nothing failing and nothing to read --
it just sat in `running`.
@node(resources={"cpus": 2}) is the declaration. The engine holds that much
for the length of the execution, so more of them than the machine has room
for wait their turn rather than oversubscribing it, and a `gpus` node holds
its card exclusively. FLOW_CPUS defaults to every core but two, and those two
are what keeps the engine answering.
Because a thread cap is read when the process imports the library, a warm
worker cannot be told a different one -- so an environment gets a pool of its
own and nodes deriving the same one share it, rather than paying a cold start
per call on exactly the nodes whose imports are slowest. XLA_FLAGS is never
derived: it is a composed, version-dependent string, so it travels in
resources.env where it is visible.
A node that declares nothing is not accounted for and behaves as it always
did -- it just gets FLOW_CPUS/FLOW_MAX_WORKERS as a thread cap, which is the
half of this that fixes the reported incident without anybody declaring
anything. An operator who set OMP_NUM_THREADS themselves still wins.
Resources are claimed strictly before a worker slot, so the two blocking
waits cannot deadlock. A node queued for them publishes node_queued and shows
on GET /workers/resources, because waiting and hanging looked identical.
Accounted, not enforced: no cgroups, no rlimits. Scheduling across machines,
flavours and enforcement are the next steps.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Fluksio App
Fluksio is a node-based automation software that brings trust and reliability to your flow. It just works and looks good.
The core of Fluksio: a node-based, test-driven automation software built to scale. This
repo holds the FastAPI backend, the flow engine, and the dashboard SPA. It is served on
app.${DOMAIN} (SPA) and api.${DOMAIN} (API)
Layout
backend/ FastAPI + SQLModel + Alembic + SQLite
fluksio/flow/ the flow engine (nodes, pipeline, state backends, controller)
fluksio/cli.py `fluksio serve` / `enroll` / `worker`
worker/ the `fluksio-worker` distribution: the agent and the node runner
frontend/ React 19 + TanStack Router + Tailwind 4 + shadcn/ui
docs/ the public documentation site (zensical), served on docs.${DOMAIN}
docker/ compose.yml → compose.dev.yml → compose.local.yml (+ compose.traefik.yml)
scripts/ generate-client.sh, test.sh
Install without Docker
Get started quickly by running
pip install fluksio
fluksio serve
Then, head over to fluksio.com, sign up and add a new installation. Using the code provided, run
fluksio enroll <code>
and you're ready to rock.
Fluksio is distributed at it's heart. A machine that should only run nodes for an engine elsewhere installs less:
pip install fluksio-worker
fluksio-worker --url wss://api.example.com/api/v1/workers/attach --token "$TOKEN" --labels gpu
Getting started
Normally driven from the workspace root (make init once, then make dev). Standalone:
make install # uv sync + bun install
make dev-utils # proxy and mailcatcher only
make dev-backend # FastAPI on :8000, hot reload
make dev-frontend # Vite on :5173
make test # pytest + Playwright (the e2e half needs the stack up)
make lint # ruff + mypy + biome
make generate-client # regenerate the frontend SDK from the OpenAPI schema
make help lists every target.
Documentation
The public site lives in docs/ and is built with zensical:
make docs-serve # live preview on :8000
make docs # static build into ./site
It is served at docs.${DOMAIN} by the docs service in docker/compose.yml,
and .gitea/workflows/docs.yml builds it with --strict on every push.
License
Copyright (C) 2026 Melvin Strobl - GNU Affero General Public License v3.0 or later. See LICENSE.