fix(docker): the slow-ingest line names which part of the Docker persist took the time (#5181)

persist_host_docker marks docker.status, .swarm, .disk, .logs, .metrics,
.events, .containers, .alerts and .reap on the ingest's phase timer, each
flushed before its mark so a part's writes are charged to it.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

plugins/docker/ingest.py

@@ -289,7 +289,7 @@ async def _expire_swarm(session, host) -> None:
 
 
 async def persist_host_docker(session, host, snapshots, swarm=None, disk=None,
-                              logs=None, status=None) -> None:
+                              logs=None, status=None, phases=None) -> None:
     """Upsert containers + time-series + lifecycle events + swarm for one host.
 
     `snapshots` is a list of (recorded_at: datetime, containers: list[dict]) —
@@ -305,20 +305,33 @@ async def persist_host_docker(session, host, snapshots, swarm=None, disk=None,
     sample's {"status", "detail", "at", "swarm_manager"} (agent ≥ 1.9.0): an
     "ok" with an empty snapshot clears the host's containers, and a node that
     says it is no longer a manager loses the swarm rows it reported as one.
+
+    `phases`, when given, is marked after each part (docker.status, .swarm,
+    .disk, .logs, .metrics, .events, .containers, .alerts, .reap), each
+    flushed before its mark, so a slow ingest names the part (#5181).
     """
     from steward.core.alerts import MetricBatch
     from .models import DockerContainer, DockerEvent, DockerMetric
 
+    async def mark(name: str) -> None:
+        if phases is not None:
+            await session.flush()
+            phases.mark(f"docker.{name}")
+
     if status is not None:
         await _persist_status(session, host, status)
         if status.get("swarm_manager") is False:
             await _expire_swarm(session, host)
+        await mark("status")
     if swarm is not None:
         await _persist_swarm(session, host, swarm)
+        await mark("swarm")
     if disk is not None:
         await _persist_disk(session, host, disk)
+        await mark("disk")
     if logs:
         await _persist_logs(session, host, logs)
+        await mark("logs")
 
     if not snapshots:
         return
@@ -337,6 +350,7 @@ async def persist_host_docker(session, host, snapshots, swarm=None, disk=None,
                     mem_pct=c.get("mem_pct") or 0.0,
                     mem_usage_bytes=c.get("mem_usage_bytes") or 0,
                 ))
+    await mark("metrics")
 
     # Snapshot of stored per-container state BEFORE the upsert overwrites it —
     # used to diff lifecycle events. Skip event derivation on the host's first
@@ -356,6 +370,7 @@ async def persist_host_docker(session, host, snapshots, swarm=None, disk=None,
                 host_id=host.id, container_name=name,
                 event=event, detail=detail, at=latest_at,
             ))
+    await mark("events")
 
     # Current state + alerts from the newest snapshot only.
     batch = MetricBatch()
@@ -423,7 +438,9 @@ async def persist_host_docker(session, host, snapshots, swarm=None, disk=None,
                 value=1.0 if health == "healthy" else 0.0,
             )
 
+    await mark("containers")
     await batch.record(session)
+    await mark("alerts")
 
     # Reap containers that vanished from the host's latest listing. Without this,
     # removed one-shot containers (CI job runners, buildkit builders, codex jobs)
@@ -437,3 +454,4 @@ async def persist_host_docker(session, host, snapshots, swarm=None, disk=None,
     if seen_names:
         reap = reap.where(DockerContainer.name.notin_(seen_names))
     await session.execute(reap)
+    await mark("reap")

plugins/host_agent/routes.py

@@ -283,6 +283,7 @@ async def ingest():
                                 "docker.persist_host_samples", UserRole.admin,
                                 session, host, docker_snapshots, latest_swarm,
                                 latest_disk, docker_log_batches, docker_status,
+                                phases=phases,
                             )
                     except Exception:
                         current_app.logger.exception(

tests/integration/test_docker.py

@@ -383,6 +383,41 @@ def test_vanished_container_is_reaped(app):
     assert after == ["keep"]   # the vanished container was reaped
 
 
+@_NEEDS_DB
+def test_persist_marks_each_part_for_the_slow_ingest_line(app):
+    """Each part of the docker persist is marked, so the slow-ingest line
+    names the part its time went to (#5181)."""
+    from steward.models.hosts import Host
+
+    class _Marks:
+        def __init__(self):
+            self.names = []
+
+        def mark(self, name):
+            self.names.append(name)
+
+    persist = _persist_fn(app)
+    at = datetime(2026, 10, 6, 23, 0, 0, tzinfo=timezone.utc)
+    c = {"name": "web", "status": "running", "cpu_pct": 1.0, "mem_pct": 1.0,
+         "restart_count": 0, "exit_code": None, "oom_killed": False, "health": None}
+    logs = [(at, [{"container": "web", "stream": "stdout", "ts": None, "line": "hello"}])]
+    marks = _Marks()
+
+    async def _go():
+        async with app.db_sessionmaker() as s:
+            async with s.begin():
+                h = Host(id=str(uuid.uuid4()), name="marks", address="10.9.9.7")
+                s.add(h)
+                await s.flush()
+                await persist(s, h, [(at, [c])], None, None, logs,
+                              {"status": "ok", "detail": None, "at": at, "swarm_manager": None},
+                              phases=marks)
+
+    asyncio.run(_go())
+    assert marks.names == ["docker.status", "docker.logs", "docker.metrics", "docker.events",
+                           "docker.containers", "docker.alerts", "docker.reap"]
+
+
 @_NEEDS_DB
 def test_an_empty_snapshot_from_a_docker_that_answered_clears_the_host(app):
     """Docker answering with no containers removes the host's rows (#5152); a