feat: allow unrestricted independent job concurrency

2026-07-27 03:19:47 +00:00
parent 1a06da3984
commit 8ed8e75447
9 changed files with 182 additions and 11 deletions
@@ -22,6 +22,47 @@ from archive_control.v1 import (


 class DaemonTransportTests(unittest.IsolatedAsyncioTestCase):
+    async def test_unrelated_job_commands_execute_concurrently(self):
+        with tempfile.TemporaryDirectory() as directory:
+            root = Path(directory)
+            token = root / "token"
+            token.write_text("shared-secret", encoding="utf-8")
+            os.chmod(token, 0o600)
+            service = ServiceConfig("http://local", PurePosixPath("/api"), root)
+            config = ClientConfig(
+                "cache-1", "Cache 1", "cache", "ws://control", token,
+                root / "state.db", root / "backups", service, service,
+            )
+            probe = FilesystemProbe(root, True, True, True, True, True)
+            daemon = ArchiveClientDaemon(config, [probe, probe], [])
+            started: set[str] = set()
+            release = asyncio.Event()
+
+            async def execute(command, correlation_id, outbound):
+                started.add(command.assign_job.job.job_id)
+                await release.wait()
+
+            daemon._execute_job_command_locked = execute
+            commands = []
+            for _ in range(2):
+                command = control_pb2.Command(command_id=str(uuid4()))
+                command.assign_job.job.job_id = str(uuid4())
+                commands.append(command)
+            outbound = asyncio.Queue()
+            tasks = [
+                asyncio.create_task(
+                    daemon._execute_job_command(command, "", outbound)
+                )
+                for command in commands
+            ]
+            for _ in range(100):
+                if len(started) == 2:
+                    break
+                await asyncio.sleep(0.01)
+            self.assertEqual(len(started), 2)
+            release.set()
+            await asyncio.gather(*tasks)
+
    async def test_silent_control_connection_ends_for_reconnect(self):
        """A lost server heartbeat must not leave durable commands stranded."""

@@ -1,5 +1,6 @@
 import hashlib
 import tempfile
+import threading
 import unittest
 from pathlib import Path
 from unittest.mock import Mock, patch
@@ -38,6 +39,82 @@ class SlowRescanSyncthing(CompleteSyncthing):


 class ClientJobHappyPathTests(unittest.TestCase):
+    def test_reconnect_replay_never_duplicates_any_transfer_step(self):
+        for step in (
+            job_pb2.JOB_STEP_KIND_SOURCE_STAGE,
+            job_pb2.JOB_STEP_KIND_SYNCTHING_TRANSFER,
+            job_pb2.JOB_STEP_KIND_TARGET_MATERIALIZE,
+            job_pb2.JOB_STEP_KIND_QB_VERIFY,
+            job_pb2.JOB_STEP_KIND_STAGING_CLEANUP,
+        ):
+            with self.subTest(step=step), 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="reconnect 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="cache-1",
+                    qbittorrent=Mock(),
+                    store=store,
+                    qb_root=root,
+                    qb_api_root=Path("/downloads"),
+                    route_path=lambda _: root,
+                    syncthing_transport=Mock(),
+                    sparse_supported=True,
+                )
+                executor.assign(control_pb2.AssignJobCommand(
+                    job=definition,
+                    expected_job_revision=1,
+                    expected_last_event_sequence=0,
+                ))
+                entered = threading.Event()
+                release = threading.Event()
+
+                def execute_step(*_args):
+                    entered.set()
+                    release.wait(1)
+                    return None
+
+                executor._execute_step = Mock(side_effect=execute_step)
+                command = control_pb2.ExecuteStepCommand(
+                    job_id=definition.job_id,
+                    expected_job_revision=1,
+                    expected_last_event_sequence=1,
+                    step=step,
+                    attempt=1,
+                )
+                results: list[list[control_pb2.JobEvent]] = []
+                first = threading.Thread(
+                    target=lambda: results.append(executor.execute(command))
+                )
+                second = threading.Thread(
+                    target=lambda: results.append(executor.execute(command))
+                )
+                first.start()
+                self.assertTrue(entered.wait(1))
+                second.start()
+                release.set()
+                first.join(1)
+                second.join(1)
+
+                self.assertFalse(first.is_alive())
+                self.assertFalse(second.is_alive())
+                self.assertEqual(executor._execute_step.call_count, 1)
+                self.assertEqual(len(results), 2)
+                self.assertEqual(results[0][-1].event_id, results[1][-1].event_id)
+
    def test_capacity_guard_fails_before_data_movement(self):
        with tempfile.TemporaryDirectory() as directory:
            root = Path(directory)