fix runtime journal run lifecycle events (#3470)

In Docker production deployments, LocalSandboxProvider runs inside the
deer-flow-gateway container, so any `sandbox.mounts[].host_path` from
config.yaml is resolved against the gateway container's filesystem — not
the host machine. When the path isn't also bind-mounted into the gateway
service, the mount was silently dropped with only a WARNING log, leaving
agents reading an empty directory in production while the same config
worked under `make dev`.

Escalate the missing-host_path branch to logger.error with explicit
guidance about Docker bind mounts and docker-compose, so the failure is
hard to miss in default log configurations. Skip behaviour is preserved
to avoid breaking existing deployments.

Also clarify the misleading `VolumeMountConfig.host_path` field
description so it documents reality for both providers:

  - LocalSandboxProvider checks host_path from inside the gateway process
    (host in `make dev`, container in `make up`).
  - AioSandboxProvider (DooD) passes host_path straight to `docker -v`
    for the sandbox container, where the host Docker daemon resolves it
    from the host machine's perspective.

config.example.yaml's `sandbox.mounts` comment gets a Note: block
pointing operators at the docker-compose bind-mount requirement so the
Docker-mode gotcha is discoverable from the canonical template.

Adds a regression test that:
  - confirms missing host_path is still skipped (no behaviour break);
  - asserts an ERROR record is emitted referencing the offending paths;
  - asserts the message contains actionable Docker/gateway/docker-compose
    keywords so future refactors can't quietly downgrade it.

Refs: https://github.com/bytedance/deer-flow/issues/3244

backend/packages/harness/deerflow/config/sandbox_config.py

@@ -4,7 +4,20 @@ from pydantic import BaseModel, ConfigDict, Field
 class VolumeMountConfig(BaseModel):
     """Configuration for a volume mount."""
 
-    host_path: str = Field(..., description="Path on the host machine")
+    host_path: str = Field(
+        ...,
+        description=(
+            "Source path for the mount. Resolution depends on the active provider: "
+            "``LocalSandboxProvider`` checks this path from the gateway process — in "
+            "``make dev`` that is the host machine, but in Docker deployments "
+            "(``make up`` / docker-compose) it is the path *inside* the "
+            "``deer-flow-gateway`` container, so the host directory must also be "
+            "bind-mounted into the gateway service for the mount to take effect. "
+            "``AioSandboxProvider`` (DooD) passes this value straight to ``docker -v`` "
+            "for the sandbox container, where it is resolved by the host Docker daemon "
+            "from the host machine's perspective."
+        ),
+    )
     container_path: str = Field(..., description="Path inside the container")
     read_only: bool = Field(default=False, description="Whether the mount is read-only")
 

backend/packages/harness/deerflow/runtime/journal.py

@@ -164,7 +164,18 @@ class RunJournal(BaseCallbackHandler):
                 metadata={"caller": caller, **(metadata or {})},
             )
 
-    def on_chain_end(self, outputs: Any, *, run_id: UUID, **kwargs: Any) -> None:
+    def on_chain_end(
+        self,
+        outputs: Any,
+        *,
+        run_id: UUID,
+        parent_run_id: UUID | None = None,
+        **kwargs: Any,
+    ) -> None:
+        # Nested chain ends fire for internal graph nodes; only the root chain
+        # represents the user-visible run lifecycle.
+        if parent_run_id is not None:
+            return
         self._put(event_type="run.end", category="outputs", content=outputs, metadata={"status": "success"})
         self._flush_sync()
 

backend/packages/harness/deerflow/sandbox/local/local_sandbox_provider.py

@@ -147,7 +147,17 @@ class LocalSandboxProvider(SandboxProvider):
                             mount.container_path,
                         )
                         continue
-                    # Ensure the host path exists before adding mapping
+                    # Ensure the host path exists before adding mapping.
+                    #
+                    # ``host_path`` is resolved against the filesystem of the
+                    # process running this provider — for ``make dev`` that is
+                    # the host machine, but for ``make up`` it is the
+                    # ``deer-flow-gateway`` container, so any host path that
+                    # isn't bind-mounted into the gateway image will be missing
+                    # here. Skipping silently makes this a high-cost-to-debug
+                    # silent failure (sandbox skill / tool reads an empty dir
+                    # instead of the configured mount), so escalate to ERROR
+                    # and include actionable guidance. See #3244.
                     if host_path.exists():
                         mappings.append(
                             PathMapping(
@@ -157,10 +167,16 @@ class LocalSandboxProvider(SandboxProvider):
                             )
                         )
                     else:
-                        logger.warning(
-                            "Mount host_path does not exist, skipping: %s -> %s",
+                        logger.error(
+                            "sandbox.mounts entry %s -> %s ignored: host_path %s does not exist from the "
+                            "perspective of the gateway process. In Docker deployments (make up / docker-compose), "
+                            "this path must also be bind-mounted into the gateway container — add a matching "
+                            "volume entry under services.gateway.volumes in docker/docker-compose.yaml (and use "
+                            "the in-container path here), or run in local mode (make dev) where the gateway sees "
+                            "the host filesystem directly.",
                             mount.host_path,
                             mount.container_path,
+                            mount.host_path,
                         )
         except Exception as e:
             # Log but don't fail if config loading fails

backend/tests/test_local_sandbox_provider_mounts.py

@@ -612,6 +612,54 @@ class TestLocalSandboxProviderMounts:
 
         assert [m.container_path for m in provider._path_mappings] == ["/mnt/skills"]
 
+    def test_setup_path_mappings_logs_actionable_error_for_missing_host_path(self, tmp_path, caplog):
+        """Regression for #3244.
+
+        When ``sandbox.mounts[].host_path`` is absent from the gateway process's
+        filesystem (the typical symptom in Docker production mode: host_path is a
+        host machine path that is not bind-mounted into the gateway container),
+        the mount is still skipped — but the failure must be a hard-to-miss ERROR
+        log with explicit, actionable guidance about Docker bind mounts, not the
+        old DEBUG/WARNING that buried the silent failure.
+        """
+        skills_dir = tmp_path / "skills"
+        skills_dir.mkdir()
+        missing_host_path = tmp_path / "does-not-exist"
+
+        from deerflow.config.sandbox_config import SandboxConfig, VolumeMountConfig
+
+        sandbox_config = SandboxConfig(
+            use="deerflow.sandbox.local:LocalSandboxProvider",
+            mounts=[
+                VolumeMountConfig(host_path=str(missing_host_path), container_path="/mnt/knowledge", read_only=True),
+            ],
+        )
+        config = SimpleNamespace(
+            skills=SimpleNamespace(container_path="/mnt/skills", get_skills_path=lambda: skills_dir, use="deerflow.skills.storage.local_skill_storage:LocalSkillStorage"),
+            sandbox=sandbox_config,
+        )
+
+        with caplog.at_level("ERROR", logger="deerflow.sandbox.local.local_sandbox_provider"):
+            with patch("deerflow.config.get_app_config", return_value=config):
+                provider = LocalSandboxProvider()
+
+        # Silent-skip behaviour is preserved (no breaking change for existing deployments).
+        assert [m.container_path for m in provider._path_mappings] == ["/mnt/skills"]
+
+        # The failure must be observable at ERROR level and reference the offending paths.
+        error_records = [r for r in caplog.records if r.levelname == "ERROR"]
+        assert error_records, "expected an ERROR log when host_path is missing"
+        message = "\n".join(r.getMessage() for r in error_records)
+        assert str(missing_host_path) in message
+        assert "/mnt/knowledge" in message
+
+        # And it must include actionable Docker guidance so users don't lose hours
+        # to a silent empty-mount failure in production.
+        lowered = message.lower()
+        assert "docker" in lowered
+        assert "gateway" in lowered
+        assert "docker-compose" in lowered
+
     def test_write_file_resolves_container_paths_in_content(self, tmp_path):
         """write_file should replace container paths in file content with local paths."""
         data_dir = tmp_path / "data"

backend/tests/test_run_journal.py

@@ -179,15 +179,16 @@ class TestLifecycleCallbacks:
         assert "run.end" in types
 
     @pytest.mark.anyio
-    async def test_nested_chain_no_run_start(self, journal_setup):
-        """Nested chains (parent_run_id set) should NOT produce run.start."""
+    async def test_nested_chain_no_run_lifecycle_events(self, journal_setup):
+        """Nested chains (parent_run_id set) should NOT produce root run lifecycle events."""
         j, store = journal_setup
         parent_id = uuid4()
         j.on_chain_start({}, {}, run_id=uuid4(), parent_run_id=parent_id)
-        j.on_chain_end({}, run_id=uuid4())
+        j.on_chain_end({}, run_id=uuid4(), parent_run_id=parent_id)
         await j.flush()
         events = await store.list_events("t1", "r1")
         assert not any(e["event_type"] == "run.start" for e in events)
+        assert not any(e["event_type"] == "run.end" for e in events)
 
 
 class TestToolCallbacks:

config.example.yaml

@@ -768,8 +768,12 @@ sandbox:
   allow_host_bash: false
   # Optional: Mount additional host directories into the sandbox.
   # Each mount maps a host path to a virtual container path accessible by the agent.
+  # Note: with LocalSandboxProvider under `make up` (docker-compose), host_path is
+  # checked from inside the deer-flow-gateway container — you must also bind-mount
+  # the same directory into services.gateway.volumes in docker/docker-compose.yaml
+  # for this mount to take effect (see issue #3244).
   # mounts:
-  #   - host_path: /home/user/my-project    # Absolute path on the host machine
+  #   - host_path: /home/user/my-project    # Absolute path; see note above for Docker mode
   #     container_path: /mnt/my-project     # Virtual path inside the sandbox
   #     read_only: true                     # Whether the mount is read-only (default: false)