Surface a failing connector poll as node health and a flow issue

The poll loop remembered what it read rather than what it published, so a
value the node could not publish counted as said: the next poll skipped it,
succeeded, and health went back to ok with the port still dark. Remember it
only after inject returns, and report ok last.

A node reporting itself down is now derived into its flow's issues on read
and counted on the health summary, so the canvas marks it and Home says so.
Being down does not stop the flow, and the issue clears by itself when the
node reports well again. The repeating poll warning is logged once per
outage rather than once per tick.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01K1moruzue2kTJd3uVisgNk

backend/fluksio/api/routes/observability.py

@@ -176,10 +176,16 @@ async def read_summary(request: Request, controller: FlowControllerDep) -> Any:
     paused = set(controller.paused_flows())
     entries = list(controller.loaded.values())
     errored = [e for e in entries if e.status is NodeStatus.ERROR]
+    unhealthy = [e for e in entries if e.health == "down"]
     if quarantined:
         problems.append(f"{len(quarantined)} flow(s) quarantined")
     if errored:
         problems.append(f"{len(errored)} node(s) failed to load")
+    if unhealthy:
+        problems.append(
+            f"{len(unhealthy)} node(s) down: "
+            f"{', '.join(sorted(e.id for e in unhealthy))}"
+        )
 
     # What the canvas flags on a flow — a dependency loop, an input nothing
     # feeds — stops that flow running just as surely as a node that will not
@@ -219,7 +225,11 @@ async def read_summary(request: Request, controller: FlowControllerDep) -> Any:
             "quarantined": len(quarantined),
             "invalid": len(invalid),
         },
-        nodes={"total": len(entries), "error": len(errored)},
+        nodes={
+            "total": len(entries),
+            "error": len(errored),
+            "unhealthy": len(unhealthy),
+        },
         queue=queue,
         loop_lag=(
             watchdog.snapshot()

backend/fluksio/flow/connector.py

@@ -86,7 +86,14 @@ class ConnectorNode(Node):
             description="Seconds between polls; 0 polls never.",
         )
 
-    __slots__ = ("config", "_poll_task", "_stop_event", "_last_published", "_artifacts")
+    __slots__ = (
+        "config",
+        "_poll_task",
+        "_stop_event",
+        "_last_published",
+        "_artifacts",
+        "_down",
+    )
 
     def __init__(self, **kwargs: Any) -> None:
         super().__init__(f=self._dispatch, **kwargs)
@@ -95,6 +102,7 @@ class ConnectorNode(Node):
         self._stop_event: asyncio.Event | None = None
         self._last_published: dict[str, Any] = {}
         self._artifacts: ArtifactStore | None = None
+        self._down = False
 
     def _dispatch(self, params: dict[str, Any], **ports: Any) -> dict[str, Any] | None:
         """The scheduler's entry point. Settings are already on ``self.config``."""
@@ -171,24 +179,29 @@ class ConnectorNode(Node):
 
         Only changed ports are published: a device polled every few seconds is
         usually saying the same thing, and every publication wakes everything
-        downstream of it.
+        downstream of it. What is remembered is what was *published*, not what
+        the poll returned — a publication that raised is retried next tick
+        rather than counting as said.
         """
         while not (self._stop_event and self._stop_event.is_set()):
             try:
                 values = await self.poll()
-                self.report_health("ok")
                 changed = {
                     port: value
                     for port, value in (values or {}).items()
                     if self._last_published.get(port, object()) != value
                 }
                 if changed:
-                    self._last_published.update(changed)
                     # inject runs the graph, which is blocking work.
                     await asyncio.to_thread(self.inject, changed)
+                    self._last_published.update(changed)
+                self.report_health("ok")
+                self._down = False
             except asyncio.CancelledError:
                 break
             except Exception as exc:
-                logger.warning("Connector '%s' failed to poll: %s", self.id, exc)
+                if not self._down:
+                    logger.warning("Connector '%s' failed to poll: %s", self.id, exc)
+                self._down = True
                 self.report_health("down", f"{type(exc).__name__}: {exc}")
             await asyncio.sleep(self.config.poll_interval)

backend/fluksio/flow/controller.py

@@ -1224,8 +1224,31 @@ class FlowController:
             }
         )
 
+    def _health_issues(self, flow: str | None = None) -> list[ValidationIssue]:
+        """Nodes that are running but not working, as issues on their flow.
+
+        Not part of `self.issues`: that list is what a build found, and this is
+        what is happening now. Derived on read, so a node reporting itself well
+        again clears it with nothing to remember.
+        """
+        return [
+            ValidationIssue(
+                code="node_unhealthy",
+                message=(
+                    f"Node '{entry.id.rpartition('.')[2]}' is down: "
+                    f"{entry.health_detail or 'no detail given'}"
+                ),
+                flow=entry.flow,
+                node=entry.id,
+            )
+            for entry in self.loaded.values()
+            if entry.health == "down" and (flow is None or entry.flow == flow)
+        ]
+
     def flow_issues(self, flow: str) -> list[ValidationIssue]:
-        return [issue for issue in self.issues if not issue.flow or issue.flow == flow]
+        return [
+            issue for issue in self.issues if not issue.flow or issue.flow == flow
+        ] + self._health_issues(flow)
 
     def preview(self, name: str) -> Preview:
         """Build a flow's unpublished draft without deploying it.

backend/fluksio/flow/pipeline.py

@@ -59,6 +59,7 @@ class ValidationIssue(BaseModel):
         "unauthenticated_hook",
         "self_loop_needs_initial",
         "missing_source",
+        "node_unhealthy",
     ]
     message: str
     flow: str = ""