The editor's mount-time fitView animated from React Flow's default viewport,
which read as the graph swiping in from the corner on every open. The first
fit is instant now, later ones stay animated, and a scaleIn wrapper gives the
same entrance the brain view has — with a re-measure on completion so the
handle bounds are not stored mid-scale.
Home only renders the brain band once some flow has nodes, so a fresh install
no longer reserves a screenful of empty space above the flows card.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01HTsT1isxUjw5gtkJk8WhuA
The ring share drops from the wordmark's 0.231 to 0.18 so the disc reads as a
dot at 32px, and both the border and the gap are rounded where they are applied
— a fractional border snaps per edge and left the disc off centre. Connections
and their port dots go up with it, pulse width included, so an arriving message
never draws its edge thinner than it rests.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01HTsT1isxUjw5gtkJk8WhuA
Plumbing only: the widget-type literal and its dtype table on both sides,
the regenerated client, a curated lucide map and four stubs the renderers
are wired to. Also a latching switch and a segmented dropdown, both a
`style` on the control that already publishes and reads back, plus the
option editor a dropdown never had.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01HTsT1isxUjw5gtkJk8WhuA
The "running" tile counted paused and invalid flows as running, so it read
"5/5" beside "1 flow(s) cannot run". Its note is now additive rather than a
precedence chain, so a quarantine no longer hides the invalid count.
Adds dur() beside si(): a ms reading picks its own unit, so a slow run reads
"1.24 s" instead of "1.2k ms".
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01HTsT1isxUjw5gtkJk8WhuA
CLOUD_CONFIG_FILE defaulted to a path relative to the working directory, so the
enrolment — this installation's credential and the portal key it pinned — sat
inside the container and was lost on every rebuild, silently detaching a
connected installation. It belongs beside the OAuth key and the flow secrets,
for exactly the same reason.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XtBzdDyLsmDaF1W7DLYtYM
The app supports Python 3.10 (requires-python, and the base image is
python:3.10); datetime.UTC only exists from 3.11. Every other module already
uses timezone.utc — this was the one file out of step, and it only passed
locally because the dev venv is 3.13. The container would not import at all.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XtBzdDyLsmDaF1W7DLYtYM
A superuser-only tab that redeems a claim code and shows the link's state. It
names the account a remote session acts as, because that is the thing being
granted and the person granting it should see it spelled out.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XtBzdDyLsmDaF1W7DLYtYM
The same bundle is served by a portal under /i/{id}, so it reads its API base,
credential and router basepath from an injected config instead of the build-time
URL and localStorage. A normal installation finds no config and behaves exactly
as before; the credential deliberately never touches localStorage, since two
installations open in one browser share an origin and would overwrite each
other's session.
The websocket URL was resolving an absolute path against the base, which
discards the base's own path — harmless until the base gained one, then it
aimed the socket at the wrong host entirely.
Connection state gets a store of its own, apart from the engine's: the proxy's
503 carries {offline, last_seen}, which raises a banner naming when the
installation was last heard from and turns a failed mutation into 'not
delivered' rather than a generic error. The screen keeps its last data
underneath, since stale readings with a timestamp beat a blank page. A
reconnecting socket invalidates every query, because whatever happened while it
was down was missed.
Verified in a browser against a real hub and installation: the full UI loads
through the tunnel with no console errors, and killing the installation raises
the banner within a poll.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XtBzdDyLsmDaF1W7DLYtYM
An installation can be enrolled with a portal by redeeming a claim code, after
which it holds one authenticated websocket open and answers proxied API calls
over it. Requests are dispatched into this process's own ASGI app, so the HTTP
trigger routes flows install at runtime are visible to it, and the live flow
stream is bridged straight off the event bus.
decode_token grows the third branch its docstring anticipated: tokens signed by
the enrolled portal resolve to the local account that performed the enrolment,
verified against a JWKS pinned at that moment. With no enrolment the branch
raises immediately, so an offline installation is unchanged and untouched.
Disconnecting deletes one file, which is the entire local revocation.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XtBzdDyLsmDaF1W7DLYtYM
A record or an artifact reference was serialised into the port row, and the
panel widened until the type selects and the buttons beside them were pushed
off its edge — a checkpoint reference is 130 characters of digest, and none of
them are what you want while wiring a flow. What shows now is what the value
*is*: 'artifact · weights.json · 60B', 'record · 3 fields'. A chevron unfolds
the whole of it, wrapped, inside the panel it belongs to.
A scalar still reads as itself, and scrolls its own overflow into view when it
is longer than the room it was given. That behaviour already existed inside
the edge inspector; it moves to Common/Marquee so the panel can have it too,
and the inspector drops its own copy of the raw-JSON block along with it.
The rows are smaller for it: the type select finally fits the word 'artifact',
and a port nothing has come through on says so with a dash rather than a
sentence — nine ports of 'nothing has come through yet' is a panel of prose
about the absence of values.
The e2e check asserts both halves: that the summary is what appears, and that
the panel is still exactly 400px with the value unfolded.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AD8SfVhzXBG2nAfFcVh3iD
make seed-demo builds demo_training — prepare on the engine, a GPU-bound
train, evaluate back here — and a panel that draws the loss curve while the
training is still going. It is the session's whole argument in one flow: batch
runs with parameters and a result, a generator yielding on a declared port
rather than logging, fluksio.emit from inside a callback, artifacts carrying
the dataset and the weights between machines, and a sweep whose configs are
isolated from each other. The train node prefers its label rather than
requiring it, so it runs before a GPU box exists and says which machine and
which numeric backend it actually used.
Building it turned up two real bugs. A run waited for a worker its flow only
*preferred*, because required_labels ignored device_policy — so the example
hung on a label it did not need. And a run's seed never reached the flow, so
sweeping over seeds ran the same experiment N times; it now fills an input of
that name when the flow declares one, which is what the field looked like it
did all along.
Pressing Run on a batch flow now submits a run rather than taking the old
non-durable path — that button is the first thing anyone evaluating will press,
and it was quietly doing something else.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AD8SfVhzXBG2nAfFcVh3iD
The first cut had node code call fluksio.log_metric, which was a second,
undeclared way for data to leave a node: invisible to validation, absent from
the canvas, and stored where the graph could not see it. That is precisely the
MLflow discrepancy this framework exists to avoid, so it is gone.
A node that produces values over time is a generator. Every yield is a dict
keyed by output port, published the instant it happens — same port, same type
check, same place on the canvas as any other value — and what it returns is
its result. A port doing this declares stream: true, and a run keeps every
number one takes, so experiment tracking is a consequence of the graph rather
than an API beside it: a chart binds to a training curve the way it binds to a
temperature. fluksio.emit writes the same ports imperatively, for where a
yield cannot reach — inside a training framework's callback.
In a live flow an emission also wakes what is downstream, as a subscriber
publishing does; in a run it does not, because a run's graph is scheduled once
and mid-node cascades would leave 'finished' with nothing to mean. The
enqueued item carries no payload: the value is already in state, and one
carrying it would re-apply an old emission after the node returned.
Verified on the stack: 30 loss values arrived live on the flow socket during a
run, attributed to the node that produced them, and the same node run on the
remote worker streamed its curve back across the socket.
Also caches remote compile results per worker, so attaching a GPU box does not
put a network round trip in every rebuild.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AD8SfVhzXBG2nAfFcVh3iD
It expects worker_main.py beside it, which is true on a worker host and not in
a clone — so trying a remote worker started with copying files around. It now
looks one directory over as well.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AD8SfVhzXBG2nAfFcVh3iD
The milestone is measured on being lighter than Kedro, so make bench-startup
measures it rather than asserting it: 61 ms from submit to result against
1110 ms for kedro run on a pipeline that does the same nothing. The difference
is not orchestration, it is that nothing is booted per run — on a 510-config
sweep that is about nine minutes of pure startup that never happens.
docs/flows/runs.md is the guide: batch flows, sweeps, reporting from inside a
node, artifacts, and the two sanctioned patterns for objects that cannot be
serialized — keep them in one node, or cross at a checkpoint.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AD8SfVhzXBG2nAfFcVh3iD
The engine runs where the automations are and the GPU is somewhere else,
usually behind a different network — so the worker connects out and the engine
answers over the socket it was given. Nothing has to expose Redis, and the
same connection works through the tunnel the hosted access will use.
What travels is the protocol the local pool already speaks, so a node cannot
tell which kind of worker it is on. A node declares device: gpu and
device_policy, the label is resolved per call (a worker attaching later needs
no rebuild), and a run whose labels nothing carries waits in the queue saying
what it waits for rather than failing — submit from the couch, the GPU box
picks it up when it is switched on.
Two things had to move with it. Compiling now happens on the machine that will
run the node: a node importing torch is correct on the GPU box and a missing
module on the engine, so checking it here failed nodes that were fine. And the
artifact endpoint accepts a worker's own credential, because storing a
checkpoint is exactly what that credential is for — and only that.
Verified against the real split: the training ran on this host (its checkpoint
names the machine and a numpy the engine does not have), streamed 40 metric
points back mid-run, and the evaluate node read the checkpoint on the engine.
Cancel kills the remote training; pulling the worker fails the run in six
seconds instead of waiting out its ten-minute timeout.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AD8SfVhzXBG2nAfFcVh3iD
A checkpoint is not a message. DType.ARTIFACT carries a reference — digest,
size, media type, name — so everything on the wire stays JSON and thirty
megabytes never sit in Redis, which answers the vision's open binary-payload
question by narrowing it: inline codecs would only serve payloads too small to
be worth a round trip, and nothing asks for that.
The store is content-addressed rather than per-run, for three reasons that all
pay later: a sweep whose fifty configs share one preprocessed input stores it
once, a reference stays valid however it is passed around because it names
content instead of a location, and the digest is what a stage cache will
compare — so building it in now is what keeps that from being a change to the
message contract.
Node code calls fluksio.save_artifact/load_artifact and cannot tell whether it
is writing the engine's own directory or putting bytes over HTTP, which is
what will let the same flow run on a remote worker unchanged.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AD8SfVhzXBG2nAfFcVh3iD
A training loop has numbers worth keeping thousands of steps before it has a
result. Node code now imports fluksio and calls log_metric/progress, which
sends a line back without ending the call; the engine writes those to
run_metric in batches from the run's own driver rather than folding them off
the event bus, which drops what it cannot keep up with.
Two things fall out. Each event resets the worker deadline, so a node's
timeout measures silence rather than duration — which is what lets a two-hour
training keep a liveness contract instead of racing it. And the worker pool's
_running is now keyed by (run, node), so cancelling one config of a sweep
kills that training and leaves the rest alone.
Fixes a latent framing bug: read_line returned whatever a read had taken,
which was fine while a worker only ever sent one line per request and
unparseable as soon as it sent several. It now keeps the remainder.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AD8SfVhzXBG2nAfFcVh3iD
A cascade has no end worth recording; a run does. Parameters go in, the graph
executes until it drains, and the result is kept — which is what an ML
experiment is and what a CI-style job is, so both are one entity.
Each run gets a state backend namespaced to itself, so two runs of one flow
cannot overwrite each other's messages; that is a constructor argument rather
than a change to the pipeline, because every key the engine keeps already goes
through the state backend. Its record is written by the driver thread rather
than folded off the event bus, which drops what it cannot keep up with. Its
own Redis stream wakes an engine up, and from the claim onwards the database
row is the truth: redelivering hours of training because an acknowledgement
was late is not recovery, so a stale lease is what marks a run whose engine
died.
Flows gain mode: batch, which are built and validated but never activated, and
nodes gain a device label for the worker that must run them.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AD8SfVhzXBG2nAfFcVh3iD
Two things the first pass got wrong.
The Changes and "Given up on" lists had no cap, so they ran as long as the
audit trail did — and worse, a truncating flex row still offers its whole
unwrapped line as a min-content contribution, so the card sized itself to the
longest entry and took the page sideways with it. Both now use the same
capped, scrolling box the runs and failures lists already had, which fixes the
length and the width together.
Running downwards, a rank of eight nodes was two thousand pixels across
because a node box is landscape: siblings cost a rank four times as much
across as they do down it. A rank wider than two now wraps onto the ranks
below, settling over a few passes since wrapping one rank pushes what was
under it up into the room that freed. The value chip on an edge no longer
reserves its width there either — that width is spent across the screen rather
than along the flow, and the rank gap it would widen is already wider than the
chip is tall. The same flow that laid out 2040x216 is now 803x722; a chain and
a diamond are unchanged, and so is every desktop layout.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VDSXaRhvqHYNevgDGmNAto
The canvas lays itself out: a layered graph, left to right on a desktop and
top to bottom on a phone, with room reserved for the value each edge carries.
Nodes cannot be dragged and `NodeDef.position` is gone from the document —
a graph nobody can arrange is one worth keeping small, which is what keeps
flows atomic. Endpoints join the same layout, so their lanes and the
localStorage that remembered where they were dragged go too.
Mobile, per the new Responsive section of DESIGN-GUIDELINES.md: the dock caps
its width and wraps instead of running off the screen, the dashboard stacks
into one column rather than shrinking a wall panel to a fifth of its size, and
Home stops widening its grid track past the viewport. A Playwright project at
a phone's width fails the build when a screen no longer fits.
Along the way: publish is the checkmark that was already there rather than a
button that appears and disappears, with discard beside it on both the flow
and the dashboard; the brain reveals a neuron's name on the first tap; and the
port sparklines get room to breathe.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VDSXaRhvqHYNevgDGmNAto
A chart is drawn in buckets, and nothing it can show changes until the
bucket it is drawing closes — so the resolution sets the refresh rather
than a flat five-second floor. A week at quarter-hour buckets now asks
four times an hour instead of sixty, for the same picture. Leaving the
field empty follows the window; a slower rate is still honoured.
`make seed-example` builds the thing to evaluate it with: a flow that
logs a temperature to InfluxDB, a flow that answers a chart's request by
turning the window into Flux and the rows back into a series, and a
dashboard holding the chart. The reading flow declares the request as an
input with a starting value, which is how a flow says a value reaches it
from a panel rather than from a node upstream.
Axis labels keep enough decimals to stay distinct — `si` rounds to three
figures, so every tick of a chart living inside one degree read "19".
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Also the one the session settled by hand: database nodes stay transport
and credentials, and the Python nodes either side do the query building
and the answer shaping. That is what a series read mode inside the node
would have prevented, so the read-mode item goes with it.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The frame centres a single reading, which is right for a gauge and wrong
for a list.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
A chart can now ask for what it draws instead of reading the ring the
engine keeps. It publishes a request — the window and the resolution —
exactly as a slider publishes a value, and draws the series a flow
answers with. What serves the request is the flow's business, so the
widget never learns which database was behind it.
The answer says what it was computed for and one computed for another
window is ignored, so two charts on one node cost a duplicate query
rather than overwriting each other's picture. Identical requests still
in flight are asked once per tab, and the refresh has a floor under it.
The panel gains the presentation the document could already hold: the
per-series label, a unit, and a y axis that can be pinned.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
A series, record or list message declares its shape instead of riding
DType.JSON, so a widget binds a shape rather than some JSON and a wrong
binding is refused before anything runs. A list declares its item type,
which is what keeps list[float] expressible for a pipeline.
On top of that: an agenda over a list, a notification over a record, and
a dashboard alert channel that publishes engine faults as one — so a
panel can show what went wrong without a flow wiring it by hand.
Also: only None means a node published nothing, a falsy value of the
wrong shape is now the named error it always should have been; and the
gauge's readout says its size is viewBox geometry rather than type scale.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
A dependency loop is flagged on the canvas and was invisible everywhere else:
/observability/summary answered "ok" with an empty problems list while the
published flow could not run at all. It now reports the flows validation
blocks, and the brain graph carries the reason on each neuron the issue names
so the view built to find broken wiring can show it.
Node errors stay counted once, as the nodes that failed to load, and an
advisory like an unauthenticated webhook marks nothing — it is worth saying,
but the flow still runs.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XC2jX6Hdj7pxGGKzBTrbqB
The Home table drew the one sparkline without its live dot, on the grounds
that a polled rollup's right edge is the last complete slice rather than this
instant. Identical curves reading differently is the worse trade, so the dot
is back everywhere and the cell keeps room for the half that hangs past the
curve.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XC2jX6Hdj7pxGGKzBTrbqB
networkidle now falls between Home's independent fetches, so make verify was
photographing the skeletons: no name, no Admin nav, "No flows yet" and zeroed
tiles on an instance with three flows.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XC2jX6Hdj7pxGGKzBTrbqB
A dashboard edit is a draft now, and the flow canvas draws the controls the
panels actually carry, so the fixture has to publish it the same way it
already publishes the flow. The assertion is unchanged.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XC2jX6Hdj7pxGGKzBTrbqB
Both overviews carried the same toolbar twice, left-aligned, with a search
field permanently taking a row of width. One `OverviewToolbar` now serves
them: the search folds into an icon and expands again on click (Escape puts
it away and hands focus back), create is a `+`, and everything sits right of
the page. Each page keeps its own create dialog — the toolbar only renders
the trigger — so the testids the runtime spec and the capture script drive
stayed where they were.
Dashboards get the flow store's draft/publish split. The editor autosaves
`dashboard.draft.json` beside `dashboard.json`; `/view/{name}`, `bindings_for`
and `history_requirements` keep reading the published file, so a wall panel
sees an edit only once someone publishes it. `POST /dashboards/{name}/publish`
and `/discard` mirror the flow routes down to the version precondition and the
409, `GET /dashboards/{name}?draft=true` is what the editor asks for, and the
dock grows the same Publish button — which flushes a queued save first, so an
autosave in flight is not published around. Creating a dashboard still writes
the published file directly: an empty document on a panel is harmless, and it
keeps the store free of a never-published case.
"Publish all" is a checkmark in the toolbar, live only when something actually
has `has_draft`. A summary carries no version and publish needs the one it is
based on, so each document's detail is read immediately before its publish —
honest against a stale list, and no version-less backend path to maintain.
Failures are counted rather than swallowed: three of five fails says so and
names the three.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XC2jX6Hdj7pxGGKzBTrbqB
The health screen was fixed at 24 hours everywhere except its lists, which
were fixed at nothing: `failuresQueryOptions` read the newest 100 rows and
`HealthActivity` filtered them client-side, so on a busy engine the failures
list covered whatever few minutes 100 rows happened to span while the chart
beside it spanned a day — and pinning an older minute showed an empty card.
One `RangePicker` now sits on the Health heading and governs the whole block:
the tiles, the flow table, both charts and both lists. Presets are 1h / 6h /
24h / 7d — the collector prunes at `OBS_RETENTION_DAYS` (30), so a week is
behind the last one. The longer windows ask for coarser buckets, since a week
of minute rollups is ten thousand points nobody can see.
Failures get the escape hatch the runs already had: a pinned minute is asked
for with `since`/`until` rather than filtered out of what is held, and the
list itself is bound to the selected range. `RUN_DEPTH`/`EVENT_DEPTH` become
one `LIST_DEPTH`, which now buys coverage of the window on screen instead of
a fixed newest-N — narrowing the range is what makes the same rows reach the
whole of it. The "Failures (24h)" tile counts errors over the selected window
from the rollups the table is drawn from, so tile, column and chart agree.
The node-panel and edge trend curves get no picker. They are a Redis ring of
the last 120 values per message with no window to ask for, so hovering one
reveals what it actually shows — how many readings, and the span they cover.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XC2jX6Hdj7pxGGKzBTrbqB
Both were already in the tree — the four workflows under .gitea/workflows and
the mosquitto/influxdb services in docker/compose.dev.yml. The CI line says
the wiring is unproven, since nothing here has watched it run on a push.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XC2jX6Hdj7pxGGKzBTrbqB
Only the soak harness was still open here: `backend/scripts/soak.py` and
`make soak` exist with all four scenarios, so the Phase 0 item is done —
with the two caveats from NOTEPAD on the line, since the redis scenario
stops a shared container and the engine kill barely stresses redelivery.
The other M5 items (brain graph, modules, out-of-process nodes, engine
history and the health screens) were already ticked and match the code.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XC2jX6Hdj7pxGGKzBTrbqB
The trend column carried `w-full`, which made it swallow every spare pixel:
the five columns beside it huddled in the left 40% with a dead gap before the
curve. It takes a bounded third now, so the auto layout spreads the rest over
the columns that have content to show, and a long flow name still widens its
own column into the card's scroll.
`si` reaches the rollup counts it was written for — 179240 executions reads as
"179k", and the failure badge and the failures tile follow. The plain window
leaves the small ones alone: 615 is still 615. The standing counts (nodes,
flows running, queue depth) stay exact, as does everything acted on digit by
digit.
The fade ran a full 40 of the 100-unit box. In a table cell three times the
width of a panel that washed out readings still worth seeing, so it is 25 —
a hint at the left end rather than a quarter of the series.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XC2jX6Hdj7pxGGKzBTrbqB
`AlertManager.send` swallowed every delivery failure, so the alerts screen's
Test button answered 200 whatever happened — the one thing it exists for. It
takes `raise_on_error` now, which only the test route passes; the per-channel
loop keeps the swallow, because one dead channel must not stop the others
hearing about the same fault. A refused delivery answers 502 with whatever the
sender said.
Renaming a flow left its values under the old name for good: the delete path
already swept them, the rename path never did. It calls the same `forget_flow`,
which covers the messages and the `__ts__`/`__version__`/`__history__`
bookkeeping keyed by message name. Cleanup, not migration — they repopulate
under the new name on the next run.
Triggering a node by hand ran `Node.__call__` with nothing catching it, so a
node that raised produced a 500 and a stack trace in the server log, and
nothing at all on the canvas. `Pipeline.publish_error` is the reporting half of
`_execute_node` lifted out; both paths go through it, so a manual failure now
reads the same on the canvas and in the metrics as a queued one. The route
answers 400 with the node's error.
`MemoryWorkQueue.stats()` counts claimed-but-unacknowledged work rather than
reporting zero, so the health tile means something without Redis. The metrics
collector's held tracebacks are capped at `DETAIL_CAP` and swept on the same
`RUN_STALE_S` cutoff the open runs use, instead of one untruncated traceback
per node kept for the life of the process — a traceback still survives the
flush between the log and the failure it belongs to.
`GET /observability/events` takes `since`/`until`, the window `/runs` already
took, so a failures list can cover the span the charts beside it are drawn from.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XC2jX6Hdj7pxGGKzBTrbqB
stroblmeandClaude Opus 5