fix: advance target steps past peer events

deploy/production/lithium/compose.yaml

@@ -2,7 +2,7 @@ name: archive-control-archive
 
 services:
   archive-client:
-    image: sodium/archive-clients:v0.1.8
+    image: sodium/archive-clients:v0.1.9
     user: "1000:1000"
     restart: unless-stopped
     command: ["--config", "/etc/archive-control/client.toml"]

deploy/production/x1/compose.yaml

@@ -2,7 +2,7 @@ name: archive-control-cache
 
 services:
   archive-client:
-    image: sodium/archive-clients:v0.1.8
+    image: sodium/archive-clients:v0.1.9
     user: "1001:1001"
     restart: unless-stopped
     network_mode: host

deploy/production/x2/compose.yaml

@@ -2,7 +2,7 @@ name: archive-control-cache
 
 services:
   archive-client:
-    image: sodium/archive-clients:v0.1.8
+    image: sodium/archive-clients:v0.1.9
     user: "1001:1001"
     restart: unless-stopped
     network_mode: host

docs/deployment-and-usage.md

@@ -218,7 +218,7 @@ cache/archive routes according to policy.
 ```yaml
 services:
   archive-client:
-    image: sodium/archive-clients:v0.1.8
+    image: sodium/archive-clients:v0.1.9
     user: "1001:1001"
     restart: unless-stopped
     command: ["archive-client", "--config", "/etc/archive-control/client.toml"]

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "archive-clients"
-version = "0.1.8"
+version = "0.1.9"
 requires-python = ">=3.11"
 dependencies = ["protobuf==7.35.1", "websockets==16.0"]
 

src/archive_clients/jobs.py

@@ -120,7 +120,11 @@ class ClientJobExecutor:
         replay = self._replay(
             command.job_id, command.expected_last_event_sequence
         )
-        if replay and replay[-1].type in {
+        step_replay = [
+            event for event in replay
+            if event.progress.step == command.step
+        ]
+        if step_replay and step_replay[-1].type in {
             control_pb2.JOB_EVENT_TYPE_STEP_SUCCEEDED,
             control_pb2.JOB_EVENT_TYPE_COMMITTED,
             control_pb2.JOB_EVENT_TYPE_SUCCEEDED,
@@ -132,31 +136,45 @@ class ClientJobExecutor:
                 for event in replay:
                     event_callback(event)
             return replay
-        started = replay[0] if replay else self._event(
-            definition,
-            sequence=command.expected_last_event_sequence + 1,
-            revision=command.expected_job_revision + 1,
-            event_type=control_pb2.JOB_EVENT_TYPE_STEP_STARTED,
-            state=job_pb2.JOB_STATE_RUNNING,
-            committed=(
-                self._committed(command.job_id)
-                or command.step == job_pb2.JOB_STEP_KIND_STAGING_CLEANUP
-                or (
-                    command.step == job_pb2.JOB_STEP_KIND_SAFE_FILE_UNLINK
-                    and self.store.get_job_artifact(
-                        command.job_id, "qb-entry-removed"
-                    ) is not None
-                )
-            ),
-            step=command.step,
-            step_state=job_pb2.STEP_STATE_RUNNING,
-        )
-        if not replay:
+        if step_replay:
+            started = step_replay[0]
+            cursor = step_replay[-1]
+            emitted = list(step_replay)
+        else:
+            previous = replay[-1] if replay else None
+            started = self._event(
+                definition,
+                sequence=(
+                    previous.sequence + 1
+                    if previous is not None
+                    else command.expected_last_event_sequence + 1
+                ),
+                revision=(
+                    previous.job_revision + 1
+                    if previous is not None
+                    else command.expected_job_revision + 1
+                ),
+                event_type=control_pb2.JOB_EVENT_TYPE_STEP_STARTED,
+                state=job_pb2.JOB_STATE_RUNNING,
+                committed=(
+                    self._committed(command.job_id)
+                    or command.step == job_pb2.JOB_STEP_KIND_STAGING_CLEANUP
+                    or (
+                        command.step == job_pb2.JOB_STEP_KIND_SAFE_FILE_UNLINK
+                        and self.store.get_job_artifact(
+                            command.job_id, "qb-entry-removed"
+                        ) is not None
+                    )
+                ),
+                step=command.step,
+                step_state=job_pb2.STEP_STATE_RUNNING,
+            )
             self._record(definition, started)
+            cursor = started
+            emitted = [started]
         if event_callback is not None:
-            event_callback(started)
-        emitted = [started]
-        cursor = started
+            for event in emitted:
+                event_callback(event)
         speed_sample = [time.monotonic(), 0]
 
         def progress(

tests/test_jobs.py

@@ -111,6 +111,72 @@ class ClientJobHappyPathTests(unittest.TestCase):
                 syncthing=SlowRescanSyncthing(),
             )
 
+    def test_target_step_advances_past_replayed_source_completion(self):
+        with tempfile.TemporaryDirectory() as directory:
+            root = Path(directory)
+            store = ClientStore(root / "client.db")
+            store.initialize()
+            definition = job_pb2.JobDefinition(
+                job_id=str(uuid4()),
+                idempotency_key=str(uuid4()),
+                operation=job_pb2.JOB_OPERATION_ARCHIVE,
+                resource_display_name="fixture",
+                transfer={
+                    "source_client_id": "cache-1",
+                    "target_client_id": "archive-1",
+                    "route_id": "route-1",
+                },
+            )
+            definition.resource_id.info_hash_v1_hex = "a" * 40
+            definition.created_at.GetCurrentTime()
+            executor = ClientJobExecutor(
+                client_id="archive-1",
+                qbittorrent=Mock(),
+                store=store,
+                qb_root=root,
+                qb_api_root=Path("/downloads"),
+                route_path=lambda _: root,
+                syncthing_transport=Mock(),
+                sparse_supported=True,
+                poll_interval=0,
+            )
+            executor.assign(control_pb2.AssignJobCommand(
+                job=definition,
+                expected_job_revision=1,
+                expected_last_event_sequence=1,
+            ))
+            source_complete = executor._event(
+                definition,
+                sequence=5,
+                revision=4,
+                event_type=control_pb2.JOB_EVENT_TYPE_STEP_SUCCEEDED,
+                state=job_pb2.JOB_STATE_RUNNING,
+                committed=False,
+                step=job_pb2.JOB_STEP_KIND_SOURCE_STAGE,
+                step_state=job_pb2.STEP_STATE_SUCCEEDED,
+            )
+            executor._record(definition, source_complete)
+
+            with patch.object(executor, "_wait_for_syncthing"):
+                events = executor.execute(control_pb2.ExecuteStepCommand(
+                    job_id=definition.job_id,
+                    expected_job_revision=4,
+                    expected_last_event_sequence=5,
+                    step=job_pb2.JOB_STEP_KIND_SYNCTHING_TRANSFER,
+                    attempt=1,
+                ))
+
+            self.assertEqual(events[0].sequence, 6)
+            self.assertEqual(
+                events[0].progress.step,
+                job_pb2.JOB_STEP_KIND_SYNCTHING_TRANSFER,
+            )
+            self.assertEqual(events[-1].sequence, 7)
+            self.assertEqual(
+                events[-1].type,
+                control_pb2.JOB_EVENT_TYPE_STEP_SUCCEEDED,
+            )
+
     def _run_transfer(
         self,
         root: Path,