Implementation of a Primary/Archive setup for object storage.

dev/environment

@@ -31,6 +31,7 @@ HIBP_API_KEY="something-not-real"
 DOCS_URL="https://pythonhosted.org/{project}/"
 
 FILES_BACKEND=warehouse.packaging.services.LocalFileStorage path=/var/opt/warehouse/packages/ url=http://localhost:9001/packages/{path}
+ARCHIVE_FILES_BACKEND=warehouse.packaging.services.LocalArchiveFileStorage path=/var/opt/warehouse/packages-archive/ url=http://localhost:9001/packages-archive/{path}
 SIMPLE_BACKEND=warehouse.packaging.services.LocalSimpleStorage path=/var/opt/warehouse/simple/ url=http://localhost:9001/simple/{path}
 DOCS_BACKEND=warehouse.packaging.services.LocalDocsStorage path=/var/opt/warehouse/docs/
 SPONSORLOGOS_BACKEND=warehouse.admin.services.LocalSponsorLogoStorage path=/var/opt/warehouse/sponsorlogos/

docker-compose.yml

@@ -3,6 +3,7 @@ version: '3.9'
 volumes:
   simple:
   packages:
+  packages-archive:
   sponsorlogos:
   policies:
   vault:
@@ -98,6 +99,7 @@ services:
       - ./pyproject.toml:/opt/warehouse/src/pyproject.toml:z
       - .coveragerc:/opt/warehouse/src/.coveragerc:z
       - packages:/var/opt/warehouse/packages
+      - packages-archive:/var/opt/warehouse/packages-archive
       - sponsorlogos:/var/opt/warehouse/sponsorlogos
       - policies:/var/opt/warehouse/policies
       - simple:/var/opt/warehouse/simple
@@ -128,6 +130,7 @@ services:
     command: python -m http.server 9001
     volumes:
       - packages:/var/opt/warehouse/packages
+      - packages-archive:/var/opt/warehouse/packages-archive
       - sponsorlogos:/var/opt/warehouse/sponsorlogos
       - simple:/var/opt/warehouse/simple
     ports:
@@ -139,10 +142,13 @@ services:
     command: hupper -m celery -A warehouse worker -B -S redbeat.RedBeatScheduler -l info
     volumes:
       - ./warehouse:/opt/warehouse/src/warehouse:z
+      - packages:/var/opt/warehouse/packages
+      - packages-archive:/var/opt/warehouse/packages-archive
     env_file: dev/environment
     environment:
       C_FORCE_ROOT: "1"
       FILES_BACKEND: "warehouse.packaging.services.LocalFileStorage path=/var/opt/warehouse/packages/ url=http://files:9001/packages/{path}"
+      ARCHIVE_FILES_BACKEND: "warehouse.packaging.services.LocalArchiveFileStorage path=/var/opt/warehouse/packages-archive/ url=http://files:9001/packages-archive/{path}"
       SIMPLE_BACKEND: "warehouse.packaging.services.LocalSimpleStorage path=/var/opt/warehouse/simple/ url=http://files:9001/simple/{path}"
 
   static:

pyproject.toml

@@ -21,6 +21,7 @@ exclude = ["warehouse/locale/.*", "warehouse/migrations/versions.*"]
 module = [
     "automat.*",
     "bpython.*",  # https://github.com/bpython/bpython/issues/892
+    "b2sdk.*",  # https://github.com/Backblaze/b2-sdk-python/issues/148
     "celery.app.backends.*",
     "celery.backends.redis.*",
     "citext.*",

requirements/main.in

@@ -1,6 +1,7 @@
 alembic>=0.7.0
 Automat
 argon2-cffi
+b2sdk
 Babel
 bcrypt
 boto3

requirements/main.txt

@@ -39,6 +39,10 @@ argon2-cffi-bindings==21.2.0 \
     --hash=sha256:f1152ac548bd5b8bcecfb0b0371f082037e47128653df2e8ba6e914d384f3c3e \
     --hash=sha256:f9f8b450ed0547e3d473fdc8612083fd08dd2120d6ac8f73828df9b7d45bb351
     # via argon2-cffi
+arrow==1.2.3 \
+    --hash=sha256:3934b30ca1b9f292376d9db15b19446088d12ec58629bc3f0da28fd55fb633a1 \
+    --hash=sha256:5a49ab92e3b7b71d96cd6bfcc4df14efefc9dfa96ea19045815914a6ab6b1fe2
+    # via b2sdk
 asn1crypto==1.5.1 \
     --hash=sha256:13ae38502be632115abf8a24cbe5f4da52e3b5231990aff31123c805306ccb9c \
     --hash=sha256:db4e40728b728508912cbb3d44f19ce188f218e9eba635821bb4b68564f8fd67
@@ -55,6 +59,9 @@ automat==22.10.0 \
     --hash=sha256:c3164f8742b9dc440f3682482d32aaff7bb53f71740dd018533f9de286b64180 \
     --hash=sha256:e56beb84edad19dcc11d30e8d9b895f75deeb5ef5e96b84a467066b3b84bb04e
     # via -r requirements/main.in
+b2sdk==1.20.0 \
+    --hash=sha256:b394d9fbdada1a4ffc0837cd6c930351f5fccc24cd0af23e41edd850d67fb687
+    # via -r requirements/main.in
 babel==2.12.1 \
     --hash=sha256:b4246fb7677d3b98f501a39d43396d3cafdc8eadb045f4a31be01863f655c610 \
     --hash=sha256:cc2d99999cd01d44420ae725a21c9e3711b3aadc7976d6147f622d8581963455
@@ -843,6 +850,10 @@ limits==3.3.1 \
     --hash=sha256:df8685b1aff349b5199628ecdf41a9f339a35233d8e4fcd9c3e10002e4419b45 \
     --hash=sha256:dfc59ed5b4847e33a33b88ec16033bed18ce444ce6a76287a4e054db9a683861
     # via -r requirements/main.in
+logfury==1.0.1 \
+    --hash=sha256:130a5daceab9ad534924252ddf70482aa2c96662b3a3825a7d30981d03b76a26 \
+    --hash=sha256:b4f04be1701a1df644afc3384d6167d64c899f8036b7eefc3b6c570c6a9b290b
+    # via b2sdk
 lxml==4.9.2 \
     --hash=sha256:01d36c05f4afb8f7c20fd9ed5badca32a2029b93b1750f571ccc0b142531caf7 \
     --hash=sha256:04876580c050a8c5341d706dd464ff04fd597095cc8c023252566a8826505726 \
@@ -1327,6 +1338,7 @@ python-dateutil==2.8.2 \
     --hash=sha256:0123cacc1627ae19ddf3c27a5de5bd67ee4586fbdd6440d9748f8abb483d3e86 \
     --hash=sha256:961d03dc3453ebbc59dbdea9e4e11c5651520a876d0f4db161e8674aae935da9
     # via
+    #   arrow
     #   botocore
     #   celery-redbeat
     #   elasticsearch-dsl
@@ -1358,6 +1370,7 @@ requests==2.28.2 \
     --hash=sha256:98b1b2782e3c6c4904938b84c0eb932721069dfdb9134313beff7c83c2df24bf
     # via
     #   -r requirements/main.in
+    #   b2sdk
     #   datadog
     #   forcediphttpsadapter
     #   google-api-core
@@ -1475,6 +1488,10 @@ text-unidecode==1.3 \
     --hash=sha256:1311f10e8b895935241623731c2ba64f4c455287888b18189350b67134a822e8 \
     --hash=sha256:bad6603bb14d279193107714b288be206cac565dfa49aa5b105294dd5c4aab93
     # via python-slugify
+tqdm==4.65.0 \
+    --hash=sha256:1871fb68a86b8fb3b59ca4cdd3dcccbc7e6d613eeed31f4c332531977b89beb5 \
+    --hash=sha256:c4f53a17fe37e132815abceec022631be8ffe1b9381c2e6e30aa70edc99e9671
+    # via b2sdk
 transaction==3.1.0 \
     --hash=sha256:65d0b1ea92dbe7c4e3b237fb6bd8b41dea23d7459e7bdd8c3880bffdaf912fa4 \
     --hash=sha256:8376a959aa71821df1bdd7d066858a3f9f34b7f5f1c0a0e1efbd11d626895449

tests/conftest.py

@@ -89,6 +89,7 @@ def metrics_timing(*args, **kwargs):
 def metrics():
     return pretend.stub(
         event=pretend.call_recorder(lambda *args, **kwargs: None),
+        gauge=pretend.call_recorder(lambda *args, **kwargs: None),
         increment=pretend.call_recorder(lambda *args, **kwargs: None),
         histogram=pretend.call_recorder(lambda *args, **kwargs: None),
         timing=pretend.call_recorder(lambda *args, **kwargs: None),
@@ -173,6 +174,13 @@ def pyramid_request(pyramid_services, jinja, remote_addr):
 
     dummy_request.registry.registerUtility(jinja, IJinja2Environment, name=".jinja2")
 
+    dummy_request._task_stub = pretend.stub(
+        delay=pretend.call_recorder(lambda *a, **kw: None)
+    )
+    dummy_request.task = pretend.call_recorder(
+        lambda *a, **kw: dummy_request._task_stub
+    )
+
     def localize(message, **kwargs):
         ts = TranslationString(message, **kwargs)
         return ts.interpolate()
@@ -264,6 +272,7 @@ def app_config(database):
         "ratelimit.url": "memory://",
         "elasticsearch.url": "https://localhost/warehouse",
         "files.backend": "warehouse.packaging.services.LocalFileStorage",
+        "archive_files.backend": "warehouse.packaging.services.LocalArchiveFileStorage",
         "simple.backend": "warehouse.packaging.services.LocalSimpleStorage",
         "docs.backend": "warehouse.packaging.services.LocalDocsStorage",
         "sponsorlogos.backend": "warehouse.admin.services.LocalSponsorLogoStorage",
@@ -273,6 +282,7 @@ def app_config(database):
             "warehouse.malware.services.PrinterMalwareCheckService"
         ),
         "files.url": "http://localhost:7000/",
+        "archive_files.url": "http://localhost:7000/archive",
         "sessions.secret": "123456",
         "sessions.url": "redis://localhost:0/",
         "statuspage.url": "https://2p66nmmycsj3.statuspage.io",

tests/unit/forklift/test_legacy.py

@@ -48,7 +48,10 @@
     Release,
     Role,
 )
-from warehouse.packaging.tasks import update_bigquery_release_files
+from warehouse.packaging.tasks import (
+    sync_file_to_archive,
+    update_bigquery_release_files,
+)
 from warehouse.utils.security_policy import AuthenticationMethod
 
 from ...common.db.accounts import EmailFactory, UserFactory
@@ -1426,17 +1429,12 @@ def storage_service_store(path, file_path, *, meta):
             "warehouse.release_files_table": "example.pypi.distributions"
         }
 
-        update_bigquery = pretend.stub(
-            delay=pretend.call_recorder(lambda *a, **kw: None)
-        )
-        db_request.task = pretend.call_recorder(lambda *a, **kw: update_bigquery)
-
         resp = legacy.file_upload(db_request)
 
         assert resp.status_code == 200
         assert db_request.find_service.calls == [
             pretend.call(IMetricsService, context=None),
-            pretend.call(IFileStorage),
+            pretend.call(IFileStorage, name="primary"),
         ]
         assert len(storage_service.store.calls) == 2 if has_signature else 1
         assert storage_service.store.calls[0] == pretend.call(
@@ -1508,7 +1506,10 @@ def storage_service_store(path, file_path, *, meta):
             )
         ]
 
-        assert db_request.task.calls == [pretend.call(update_bigquery_release_files)]
+        assert db_request.task.calls == [
+            pretend.call(update_bigquery_release_files),
+            pretend.call(sync_file_to_archive),
+        ]
 
         assert metrics.increment.calls == [
             pretend.call("warehouse.upload.attempt"),
@@ -2567,6 +2568,47 @@ def test_upload_fails_with_unsafe_filename(
         assert resp.status_code == 400
         assert resp.status == "400 Cannot upload a file with '/' or '\\' in the name."
 
+    @pytest.mark.parametrize("character", [*(chr(x) for x in range(32)), chr(127)])
+    def test_upload_fails_with_disallowed_in_filename(
+        self, pyramid_config, db_request, character
+    ):
+
+        user = UserFactory.create()
+        pyramid_config.testing_securitypolicy(identity=user)
+        db_request.user = user
+        EmailFactory.create(user=user)
+        project = ProjectFactory.create()
+        release = ReleaseFactory.create(project=project, version="1.0")
+        RoleFactory.create(user=user, project=project)
+
+        filename = f"{project.name}{character}-{release.version}.tar.wat"
+
+        db_request.POST = MultiDict(
+            {
+                "metadata_version": "1.2",
+                "name": project.name,
+                "version": release.version,
+                "filetype": "sdist",
+                "md5_digest": "nope!",
+                "content": pretend.stub(
+                    filename=filename,
+                    file=io.BytesIO(b"a" * (legacy.MAX_FILESIZE + 1)),
+                    type="application/tar",
+                ),
+            }
+        )
+
+        with pytest.raises(HTTPBadRequest) as excinfo:
+            legacy.file_upload(db_request)
+
+        resp = excinfo.value
+
+        assert resp.status_code == 400
+        assert resp.status == (
+            "400 Cannot upload a file with non-printable characters (ordinals 0-31) "
+            "or the DEL character (ordinal 127) in the name."
+        )
+
     def test_upload_fails_without_user_permission(self, pyramid_config, db_request):
         user1 = UserFactory.create()
         EmailFactory.create(user=user1)
@@ -2786,7 +2828,7 @@ def storage_service_store(path, file_path, *, meta):
         assert resp.status_code == 200
         assert db_request.find_service.calls == [
             pretend.call(IMetricsService, context=None),
-            pretend.call(IFileStorage),
+            pretend.call(IFileStorage, name="primary"),
         ]
         assert storage_service.store.calls == [
             pretend.call(
@@ -2898,7 +2940,7 @@ def storage_service_store(path, file_path, *, meta):
         assert resp.status_code == 200
         assert db_request.find_service.calls == [
             pretend.call(IMetricsService, context=None),
-            pretend.call(IFileStorage),
+            pretend.call(IFileStorage, name="primary"),
         ]
         assert storage_service.store.calls == [
             pretend.call(

tests/unit/packaging/test_init.py

@@ -28,6 +28,7 @@
 from warehouse.packaging.models import File, Project, Release, Role
 from warehouse.packaging.services import project_service_factory
 from warehouse.packaging.tasks import (  # sync_bigquery_release_files,
+    check_file_archive_tasks_outstanding,
     compute_2fa_mandate,
     update_description_html,
 )
@@ -49,6 +50,7 @@ def key_factory(keystring, iterate_on=None, if_attr_exists=None):
     monkeypatch.setattr(packaging, "key_factory", key_factory)
     settings = {
         "files.backend": "foo.bar",
+        "archive_files.backend": "peas.carrots",
         "simple.backend": "bread.butter",
         "docs.backend": "wu.tang",
         "warehouse.packaging.project_create_user_ratelimit_string": "20 per hour",
@@ -73,7 +75,8 @@ def key_factory(keystring, iterate_on=None, if_attr_exists=None):
     packaging.includeme(config)
 
     assert config.register_service_factory.calls == [
-        pretend.call(storage_class.create_service, IFileStorage),
+        pretend.call(storage_class.create_service, IFileStorage, name="primary"),
+        pretend.call(storage_class.create_service, IFileStorage, name="archive"),
         pretend.call(storage_class.create_service, ISimpleStorage),
         pretend.call(storage_class.create_service, IDocsStorage),
         pretend.call(
@@ -169,6 +172,10 @@ def key_factory(keystring, iterate_on=None, if_attr_exists=None):
             in config.add_periodic_task.calls
         )
 
+    assert (
+        pretend.call(crontab(minute="*/1"), check_file_archive_tasks_outstanding)
+        in config.add_periodic_task.calls
+    )
     assert (
         pretend.call(crontab(minute="*/5"), update_description_html)
         in config.add_periodic_task.calls