Raw cpus and gpus are a property of the machines an installation has, so a node
written against a cluster quietly stops meaning anything when the cluster is
replaced. A node says "gpu-small" instead, and what that is stored here —
editable, and read again every time the node is built, so changing the flavor
changes what the next run gets.
Memory joins the schema properly (`ram`, in MB, accepting "2G"), along with
`duration_s` for how long a node is expected to take. That one is recorded and
shown and nothing else yet: a statement for whoever is planning around the node,
not a limit — the limit is still `timeout`.
A flavor and a number for the same thing is refused, compared by value so an
editor writing the whole object back with its defaults still round-trips. A name
nothing stores is refused at the save, which covers the canvas and `fluksio
sync` at once, and deleting one a node still asks for says which node.
Four sizes are seeded on an installation that has none, and never re-seeded:
re-adding one somebody deliberately removed is an argument nobody wins.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01A6HeySA27EkGANZN95QySW
The engine answered "where does this node run" twice, in two ways that could
not see each other: a device sent it to a worker carrying that label, and
resources were counted against the engine's own cores. Declaring both meant the
second answer won and nothing was counted at all — which the data-science
getting-started page and the worked example both do.
One question now, in flow/placement.py: of every machine attached, which could
grant what this node asked for, and which of those has it free. The books move
onto each machine — one accountant per worker, built from the inventory it
reported — and the waiting moves above them, where one condition variable can
be woken by a release anywhere or by a worker attaching. Locks go one way:
placer, then a machine's books, never back.
So a node asking for a card now finds the box that has one, rather than being
clamped down to none and run here. When nothing can grant the ask at all it is
still cut down and run — a flow written on a cluster has to work on a laptop —
but the ceiling is one real machine now, since taking the largest of each
dimension separately can describe a machine nobody has.
Two things fixed on the way. A device on a connector node held every batch run
of its flow forever, waiting for a worker that could never run an entry point.
And `prefer` falling back to the engine skipped the books, so the fallback held
nothing.
The bench flow's node has taken a `params` argument that with_settings has not
forwarded for some time, so the benchmark could not run at all: 62 ms median
submit-to-result with this, against the 61 ms on record.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01A6HeySA27EkGANZN95QySW
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>
stroblmeandClaude Opus 5