release: archive clients v0.1.18

Dockerfile

@@ -11,8 +11,12 @@ CMD ["python", "-m", "unittest", "discover", "-s", "tests", "-v"]
 
 FROM python:3.11-slim
 ENV PYTHONDONTWRITEBYTECODE=1 PYTHONUNBUFFERED=1
-RUN groupadd --gid 1001 archive-control \
+RUN if ! getent group 1001 >/dev/null; then \
-    && useradd --uid 1001 --gid 1001 --no-create-home archive-control
+        groupadd --gid 1001 archive-control; \
+    fi \
+    && if ! getent passwd 1001 >/dev/null; then \
+        useradd --uid 1001 --gid 1001 --no-create-home archive-control; \
+    fi
 COPY --from=builder /wheels /wheels
 RUN pip install --no-cache-dir /wheels/*.whl && rm -rf /wheels
 USER 1001:1001

deploy/production/lithium/compose.yaml

@@ -2,7 +2,7 @@ name: archive-control-archive
 
 services:
   archive-client:
-    image: sodium/archive-clients:v0.1.13
+    image: sodium/archive-clients:v0.1.14
     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:v0.1.13
+    image: sodium/archive-clients:v0.1.14
     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:v0.1.13
+    image: sodium/archive-clients:v0.1.14
     user: "1001:1001"
     restart: unless-stopped
     network_mode: host

docs/decisions.md

@@ -35,8 +35,11 @@ expand these rules but must not contradict them.
   manifests, and observed external state.
 - SQLite backups use the online backup API. Defaults are every six hours, 12
   recent, 14 daily, and eight weekly copies, plus pre/post-migration backups.
-- Default concurrency is one active data-moving job per client and per route.
+- Default concurrency is unrestricted: every eligible queued job is admitted
-  Disjoint node pairs may run concurrently. Queueing is durable FIFO.
+  in one scheduler pass and independent jobs execute concurrently on a client.
+  Physical qBittorrent, Syncthing, disk, and network capacity are therefore
+  the natural limit. `enforce_concurrency_limits=true` restores the optional
+  legacy per-client/per-route gates. Commands for one job remain serialized.
 - A queued job owns a per-resource reservation. Cancelling it removes only the
   queued record/reservation and never sends cleanup commands.
 - Connectivity or transfer stalls wait indefinitely. A configurable 30-minute

docs/deployment-and-usage.md

@@ -44,6 +44,9 @@ command_max_attempts = 3 # Total sends, including the initial attempt.
 stall_after = "30m" # Warning state only; jobs continue waiting.
 route_policy = "on_demand" # Alternative: eager_mesh.
 route_setup_timeout = "30m"
+# Default: allow all eligible jobs; disk/network/service capacity is the limit.
+enforce_concurrency_limits = false
+# Used only when enforce_concurrency_limits is true.
 max_active_per_client = 1
 max_active_per_route = 1
 max_envelope_bytes = 1048576
@@ -223,7 +226,7 @@ cache/archive routes according to policy.
 ```yaml
 services:
   archive-client:
-    image: sodium/archive-clients:v0.1.13
+    image: sodium/archive-clients:v0.1.14
     user: "1001:1001"
     restart: unless-stopped
     command: ["archive-client", "--config", "/etc/archive-control/client.toml"]
@@ -256,6 +259,28 @@ Generated protobuf Python bindings are committed in each consumer with the
 exact `archive-control-proto` tag/commit recorded. Release order is proto,
 control consumer, client consumer, E2E, then image publication.
 
+### Reproducible multi-platform Buildx lifecycle
+
+The named Buildx builders are local acceleration/cache only; they are not a
+deployment dependency and can be removed after publication. To create a fresh
+builder, verify its platforms, publish a release, and remove it afterwards:
+
+```bash
+docker buildx create --name archive-control-release --driver docker-container --use
+docker buildx inspect --bootstrap
+docker buildx build --platform linux/amd64,linux/arm64 \
+  --tag sodium/archive-clients:vX.Y.Z --push .
+docker buildx rm archive-control-release
+```
+
+`docker buildx inspect` must list both `linux/amd64` and `linux/arm64` before
+publishing. If the host has no arm64 emulation, install/configure it according
+to the host Docker distribution before the build; do not publish a partial
+single-platform tag. Retain the pushed manifest digest in the release notes
+and deploy the immutable tag or digest. The optional `archive-control-qemu`
+builder follows the same lifecycle when it is used for an emulation smoke
+build.
+
 ## Operator usage
 
 1. Prepare local qB, sync, state, backup, config, and secret mounts with the

docs/system-design.md

@@ -133,10 +133,13 @@ a resource may be active or queued at a time. This prevents incompatible
 baselines even when jobs would use different nodes or observe different sides
 of a hybrid identity.
 
-Defaults allow one active data-moving job per client and one per route. A job
+By default every eligible queued job is claimed in the same scheduler pass and
-must acquire its source client, target client, route, and resource reservation
+independent jobs run concurrently on each client; physical service and storage
-atomically. Disjoint node pairs may run concurrently. Route setup is a
+capacity are the limit. Set `enforce_concurrency_limits=true` on control to
-preflight activity and does not permit a data step to bypass these leases.
+enable the optional one-per-client/one-per-route gates. A job always retains
+its resource reservation, and commands for the same job remain serialized.
+Route setup is a preflight activity and does not permit a data step to bypass
+the resource reservation.
 
 Offline nodes do not prevent unrelated jobs from being listed or run. A job
 requiring an offline node remains waiting indefinitely; it does not consume an
@@ -154,6 +157,12 @@ advances to the next step. On reconnect:
 4. Control observes relevant qBittorrent, Syncthing, staging, and manifest
    state before selecting retry, resume, compensation, cleanup, or manual
    intervention.
+
+If a connection disappears while a synchronous job operation is in progress,
+the client journals events locally, reconnects indefinitely, and serializes a
+replayed command behind the in-flight operation for that job. The replay reads
+the journal rather than repeating the data operation. This applies to every
+transfer step, including post-commit staging cleanup.
 5. A command is reissued with its original ID when the acceptance result is
    uncertain.
 

docs/telegram-ux.md

@@ -25,8 +25,10 @@ What do you want to do?
 [ Evict Cache ] [ Job Status ]
 ```
 
-All subsequent pages edit this message. `Cancel` closes the active selection
+All subsequent pages edit this message. Leaving a resource/tree selection
-flow. `Back` returns one level while retaining validated filters/selections.
+returns to the Archive Control operation menu without closing the shared
+conversation. Confirmation `Cancel` returns to its immediate prior selection;
+the main menu alone can close Archive Control.
 Callback payloads contain opaque session/action IDs, not resource names or
 paths, and are validated against persisted session revision and expiry.
 

docs/testing.md

@@ -225,7 +225,7 @@ Fixtures include small deterministic v1, v2, and hybrid torrents with:
 | Recovery | restart every step on source/target/control, lost DB with proof, ambiguous loss fails closed |
 | Messaging | duplicate command/event, lost ack, sequence gap, stale revision, duplicate client ID |
 | Safety | attempted download, corrupt same-size file, symlink/path escape, special file, partfile mismatch |
-| Operations | stall retains percent, offline wait, cancellation each phase, postcommit cleanup retry |
+| Operations | stall retains percent, offline wait, cancellation each phase, postcommit cleanup retry, disconnect/reconnect during every transfer step with exactly-once replay |
 | Database | online backup under load, pre/post migration, retention, corrupt backup rejection, offline restore |
 | UI | pagination/filter chain, stale/double callback, persisted session, clear vs evict separation |
 

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "archive-clients"
-version = "0.1.13"
+version = "0.1.18"
 requires-python = ">=3.11"
 dependencies = ["protobuf==7.35.1", "websockets==16.0"]
 

src/archive_clients/daemon.py

@@ -43,6 +43,19 @@ from archive_control.v1 import (
 logger = logging.getLogger(__name__)
 
 
+def _job_id_for_command(command: control_pb2.Command) -> str:
+    """Return the durable job key for commands whose execution is serialized."""
+
+    payload = command.WhichOneof("payload")
+    if payload == "assign_job":
+        return command.assign_job.job.job_id
+    if payload == "execute_step":
+        return command.execute_step.job_id
+    if payload == "cancel_job":
+        return command.cancel_job.job_id
+    raise ValueError(f"command {command.command_id} does not execute a job")
+
+
 class ArchiveClientDaemon:
     def __init__(
         self,
@@ -95,7 +108,10 @@ class ArchiveClientDaemon:
         self._lease = DatabaseLease(config.state_db)
         self._active_route_commands: set[str] = set()
         self._active_job_commands: set[str] = set()
-        self._job_execution_lock = asyncio.Lock()
+        # Commands for one job remain ordered locally, while unrelated jobs
+        # may use the node's available qB/Syncthing/filesystem capacity in
+        # parallel. The control daemon owns admission policy.
+        self._job_execution_locks: dict[str, asyncio.Lock] = {}
         self.jobs = (
             ClientJobExecutor(
                 client_id=config.client_id,
@@ -560,7 +576,9 @@ class ArchiveClientDaemon:
         correlation_id: str,
         outbound: asyncio.Queue[str],
     ) -> None:
-        async with self._job_execution_lock:
+        job_id = _job_id_for_command(command)
+        lock = self._job_execution_locks.setdefault(job_id, asyncio.Lock())
+        async with lock:
             await self._execute_job_command_locked(
                 command, correlation_id, outbound
             )

src/archive_clients/jobs.py

@@ -86,6 +86,8 @@ class ClientJobExecutor:
         self.verification_timeout = verification_timeout
         self.free_space_reserve_bytes = free_space_reserve_bytes
         self._cancel_events: dict[str, threading.Event] = {}
+        self._execution_locks: dict[str, threading.Lock] = {}
+        self._execution_locks_guard = threading.Lock()
 
     def request_cancel(self, job_id: str) -> None:
         self._cancel_events.setdefault(job_id, threading.Event()).set()
@@ -115,6 +117,22 @@ class ClientJobExecutor:
         self,
         command: control_pb2.ExecuteStepCommand,
         event_callback: Callable[[control_pb2.JobEvent], None] | None = None,
+    ) -> list[control_pb2.JobEvent]:
+        # asyncio cancellation of a connection-bound task cannot stop the
+        # synchronous filesystem/qB operation already running in its worker
+        # thread. A replay after reconnect therefore waits for that operation
+        # and then reads its durable event journal instead of executing twice.
+        with self._execution_lock(command.job_id):
+            return self._execute_locked(command, event_callback)
+
+    def _execution_lock(self, job_id: str) -> threading.Lock:
+        with self._execution_locks_guard:
+            return self._execution_locks.setdefault(job_id, threading.Lock())
+
+    def _execute_locked(
+        self,
+        command: control_pb2.ExecuteStepCommand,
+        event_callback: Callable[[control_pb2.JobEvent], None] | None = None,
     ) -> list[control_pb2.JobEvent]:
         definition = self._definition(command.job_id)
         replay = self._replay(
@@ -653,6 +671,15 @@ class ClientJobExecutor:
                 # Syncthing can expose the per-job directory before the
                 # ready marker and manifest have arrived atomically as a set.
                 pass
+            except RouteSetupError as exc:
+                # A network interruption can temporarily make the local
+                # Syncthing REST API unavailable.  The staged payload is
+                # durable and the job must remain recoverable; keep polling
+                # instead of turning a transient outage into a failed job.
+                logger.warning("syncthing_status_deferred", extra={
+                    "job_id": definition.job_id,
+                    "error_type": type(exc).__name__,
+                })
             time.sleep(self.poll_interval)
 
     def _target_materialize(

src/archive_clients/qbittorrent.py

@@ -3,6 +3,7 @@
 from __future__ import annotations
 
 import json
+import logging
 import threading
 import time
 import uuid
@@ -16,6 +17,8 @@ from urllib import error, parse, request
 from archive_clients.config import ServiceConfig
 from archive_clients.resources import NormalizedResource, normalize_resource
 
+logger = logging.getLogger(__name__)
+
 
 class QBittorrentError(RuntimeError):
     pass
@@ -72,7 +75,17 @@ class QBittorrentReader:
             torrent_hash = torrent.get("hash")
             if not isinstance(torrent_hash, str):
                 raise QBittorrentError("qBittorrent torrent hash is invalid")
-            result.append(self._normalize(torrent, torrent_hash))
+            try:
+                result.append(self._normalize(torrent, torrent_hash))
+            except ValueError as exc:
+                # A stale or malformed qBittorrent entry must not hide every
+                # otherwise valid resource from Archive/Evict inventory.
+                # Keep it ineligible and leave an operator-visible diagnosis.
+                logger.warning(
+                    "Skipping qBittorrent resource with inconsistent "
+                    "metadata: hash=%s name=%r reason=%s",
+                    torrent_hash, torrent.get("name"), exc,
+                )
         return result
 
     def get_resource(self, torrent_hash: str) -> NormalizedResource | None:

src/archive_clients/resources.py

@@ -91,12 +91,21 @@ def normalize_resource(
     observed_at: datetime | None = None,
 ) -> NormalizedResource:
     metainfo = decode_metainfo(metainfo_bytes)
-    if len(raw_files) != len(metainfo.files):
+    # qBittorrent/libtorrent may omit torrent padding files from
+    # /torrents/files while keeping them in the exported metainfo.
+    visible_metainfo_files = tuple(
+        item for item in metainfo.files if not item.padding
+    )
+    if len(raw_files) == len(metainfo.files):
+        matched_metainfo_files = metainfo.files
+    elif len(raw_files) == len(visible_metainfo_files):
+        matched_metainfo_files = visible_metainfo_files
+    else:
         raise ResourceError("qBittorrent and metainfo file counts differ")
     files = []
     canonical = True
     for expected_index, (raw, meta_file) in enumerate(
-        zip(raw_files, metainfo.files, strict=True)
+        zip(raw_files, matched_metainfo_files, strict=True)
     ):
         index = _integer(raw.get("index"), "file index")
         if index != expected_index:

src/archive_clients/state.py

@@ -7,6 +7,8 @@ import json
 import os
 import sqlite3
 import stat
+import threading
+from contextlib import contextmanager
 from dataclasses import dataclass
 from pathlib import Path
 
@@ -14,6 +16,12 @@ from pathlib import Path
 SCHEMA_VERSION = 3
 
 
+# A client daemon executes unrelated jobs concurrently.  SQLite still permits
+# only one writer, so serialize this process's short state transactions rather
+# than allowing an otherwise healthy job to fail after its busy timeout.
+_DATABASE_LOCK = threading.RLock()
+
+
 class CommandConflict(RuntimeError):
     pass
 
@@ -557,14 +565,30 @@ class ClientStore:
             ).fetchall()
         return [dict(row) for row in rows]
 
-    def _connect(self) -> sqlite3.Connection:
+    @contextmanager
-        connection = sqlite3.connect(self.database, isolation_level=None, timeout=5)
+    def _connect(self):
-        connection.row_factory = sqlite3.Row
+        """Yield one connection while serializing local SQLite writers.
-        connection.execute("PRAGMA foreign_keys = ON")
+
-        connection.execute("PRAGMA journal_mode = WAL")
+        The longer SQLite timeout also covers a short lock held by a separate
-        connection.execute("PRAGMA synchronous = FULL")
+        maintenance process such as a backup.  The lock is deliberately held
-        connection.execute("PRAGMA busy_timeout = 5000")
+        for the full transaction, including ``BEGIN IMMEDIATE``.
-        return connection
+        """
+        with _DATABASE_LOCK:
+            connection = sqlite3.connect(
+                self.database, isolation_level=None, timeout=30
+            )
+            try:
+                connection.row_factory = sqlite3.Row
+                connection.execute("PRAGMA foreign_keys = ON")
+                connection.execute("PRAGMA journal_mode = WAL")
+                connection.execute("PRAGMA synchronous = FULL")
+                connection.execute("PRAGMA busy_timeout = 30000")
+                # Preserve the original ``with connection`` commit/rollback
+                # behavior used by every store operation.
+                with connection:
+                    yield connection
+            finally:
+                connection.close()
 
 
 def _canonical(value: object) -> str:

tests/test_daemon.py

@@ -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."""
 

tests/test_jobs.py

@@ -1,5 +1,6 @@
 import hashlib
 import tempfile
+import threading
 import unittest
 from pathlib import Path
 from unittest.mock import Mock, patch
@@ -11,7 +12,7 @@ from archive_clients.jobs import (
     JobExecutionError,
     _resource_fingerprint,
 )
-from archive_clients.syncthing import RouteSetupError
+from archive_clients.syncthing import RouteSetupError, SyncthingTransferStatus
 from archive_clients.resources import normalize_resource
 from archive_clients.state import ClientStore
 from archive_control.v1 import control_pb2, job_pb2
@@ -38,6 +39,120 @@ class SlowRescanSyncthing(CompleteSyncthing):
 
 
 class ClientJobHappyPathTests(unittest.TestCase):
+    def test_syncthing_api_outage_during_transfer_is_retried(self):
+        """A transient local REST outage must not terminally fail the job."""
+        with tempfile.TemporaryDirectory() as directory:
+            root = Path(directory)
+            store = ClientStore(root / "client.db")
+            store.initialize()
+            definition = job_pb2.JobDefinition(
+                job_id=str(uuid4()),
+                transfer={
+                    "source_client_id": "cache-1",
+                    "target_client_id": "archive-1",
+                    "route_id": "route-1",
+                },
+            )
+            observer = Mock()
+            observer.status.side_effect = [
+                RouteSetupError("Syncthing API is unavailable"),
+                SyncthingTransferStatus(1, 42, 42, True, 0),
+            ]
+            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._observer = Mock(return_value=observer)
+            progress = Mock()
+
+            executor._wait_for_syncthing(definition, progress)
+
+            self.assertEqual(observer.status.call_count, 2)
+            progress.assert_called_once_with(1, 42, 42, "0 Syncthing items still needed")
+
+    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)

tests/test_qbittorrent.py

@@ -205,6 +205,54 @@ class QBittorrentReaderTests(unittest.TestCase):
         self.assertEqual(resources, [])
         self.assertEqual(len(opener.calls), 2)
 
+    def test_listing_skips_malformed_torrent_and_keeps_valid_resources(self):
+        def torrent(name, content):
+            info = {
+                b"length": len(content), b"name": name.encode(),
+                b"piece length": 16384, b"pieces": b"x" * 20,
+            }
+            return encode({b"info": info}), hashlib.sha1(encode(info)).hexdigest()
+
+        first_bytes, first_hash = torrent("first.txt", b"one")
+        malformed_bytes, malformed_hash = torrent("broken.txt", b"bad")
+        second_bytes, second_hash = torrent("second.txt", b"two")
+        torrents = [
+            {"hash": first_hash, "name": "first.txt", "state": "uploading"},
+            {"hash": malformed_hash, "name": "broken.txt", "state": "stalledUP"},
+            {"hash": second_hash, "name": "second.txt", "state": "uploading"},
+        ]
+        responses = [
+            b"Ok.", json.dumps(torrents).encode(),
+            b'[{"index":0,"name":"first.txt","size":3,"progress":1,"priority":1}]',
+            first_bytes,
+            b'[{"index":0,"name":"broken.txt","size":3,"progress":1,"priority":1},'
+            b'{"index":1,"name":"unexpected.txt","size":1,"progress":1,"priority":1}]',
+            malformed_bytes,
+            b'[{"index":0,"name":"second.txt","size":3,"progress":1,"priority":1}]',
+            second_bytes,
+        ]
+        with tempfile.TemporaryDirectory() as directory:
+            root = Path(directory)
+            password = root / "password"
+            password.write_text("secret", encoding="utf-8")
+            os.chmod(password, 0o600)
+            config = ServiceConfig(
+                "http://qb", PurePosixPath("/downloads"), root,
+                username="admin", password_file=password,
+            )
+            opener = _Opener(responses)
+            with patch(
+                "archive_clients.qbittorrent.request.build_opener",
+                return_value=opener,
+            ), self.assertLogs("archive_clients.qbittorrent", "WARNING") as logs:
+                resources = QBittorrentReader(config).list_resources()
+
+        self.assertEqual(
+            [resource.summary.display_name for resource in resources],
+            ["first.txt", "second.txt"],
+        )
+        self.assertIn(malformed_hash, "\n".join(logs.output))
+
     def test_stopped_add_selection_recheck_and_entry_only_delete(self):
         torrent_hash = "a" * 40
         responses = [

tests/test_resources.py

@@ -118,6 +118,47 @@ class ResourceTests(unittest.TestCase):
         self.assertEqual(root.available_file_count, 1)
         self.assertEqual(root.available_logical_bytes, 3)
 
+    def test_qbittorrent_hidden_padding_files_are_normalized(self):
+        info = {
+            b"files": [
+                {b"length": 3, b"path": [b"first.bin"]},
+                {
+                    b"attr": b"p", b"length": 5,
+                    b"path": [b"_____padding_file_5"],
+                },
+                {b"length": 7, b"path": [b"last.bin"]},
+            ],
+            b"name": b"with-padding",
+            b"piece length": 16384,
+            b"pieces": b"x" * 20,
+        }
+        metainfo = encode({b"info": info})
+        torrent_hash = hashlib.sha1(encode(info)).hexdigest()
+        normalized = normalize_resource(
+            {
+                "hash": torrent_hash,
+                "name": "with-padding",
+                "state": "stalledUP",
+            },
+            [
+                {
+                    "index": 0, "name": "with-padding/first.bin",
+                    "size": 3, "completed": 3, "priority": 1,
+                },
+                {
+                    "index": 1, "name": "with-padding/last.bin",
+                    "size": 7, "completed": 7, "priority": 1,
+                },
+            ],
+            metainfo,
+        )
+        self.assertEqual(
+            [item.canonical_path for item in normalized.files],
+            ["with-padding/first.bin", "with-padding/last.bin"],
+        )
+        self.assertEqual(normalized.summary.total_file_count, 2)
+        self.assertEqual(normalized.summary.selected_complete_bytes, 10)
+
     def test_noncanonical_and_unsafe_paths_are_distinct(self):
         info = {
             b"length": 3,

tests/test_state.py

@@ -1,6 +1,7 @@
 import tempfile
 import unittest
 import os
+import threading
 from pathlib import Path
 from uuid import uuid4
 
@@ -13,6 +14,40 @@ from archive_clients.state import (
 
 
 class ClientStoreTests(unittest.TestCase):
+    def test_database_connections_serialize_local_writers(self):
+        """Concurrent jobs share one daemon DB without SQLite lock failures."""
+        with tempfile.TemporaryDirectory() as directory:
+            database = Path(directory) / "state.db"
+            first = ClientStore(database)
+            second = ClientStore(database)
+            first.initialize()
+            barrier = threading.Barrier(2)
+            failures: list[Exception] = []
+
+            def write(store, command_id):
+                try:
+                    barrier.wait()
+                    store.accept_command(command_id, '{"kind":"job"}', '{"ok":true}')
+                except Exception as exc:  # pragma: no cover - assertion below
+                    failures.append(exc)
+
+            left = threading.Thread(target=write, args=(first, "command-1"))
+            right = threading.Thread(target=write, args=(second, "command-2"))
+            left.start()
+            right.start()
+            left.join(1)
+            right.join(1)
+
+            self.assertFalse(left.is_alive())
+            self.assertFalse(right.is_alive())
+            self.assertEqual(failures, [])
+            self.assertEqual(len(first.list_accepted_commands()), 2)
+            with first._connect() as connection:
+                self.assertEqual(
+                    connection.execute("PRAGMA busy_timeout").fetchone()[0],
+                    30000,
+                )
+
     def test_command_acceptance_is_durable_and_content_addressed(self):
         with tempfile.TemporaryDirectory() as directory:
             database = Path(directory) / "state.db"