Merge branch 'releases/1.28.0'

CHANGELOG.md

@@ -16,11 +16,13 @@ These are the section headers that we use:
 
 ## [Unreleased]()
 
+## [1.28.0](https://github.com/argilla-io/argilla-server/compare/v1.27.0...v1.28.0)
+
 ### Added
 
 - Added support to specify a list of score values for suggestions `score` attribute. ([#98](https://github.com/argilla-io/argilla-server/pull/98))
 - Added `GET /api/v1/settings` new endpoint exposing Argilla and Hugging Face settings when available. ([#127](https://github.com/argilla-io/argilla-server/pull/127))
-- Added `ARGILLA_SHOW_HUGGINGFACE_SPACE_PERSISTANT_STORAGE_WARNING` new environment variable to disable warning message when Hugging Face Spaces persistent storage is disabled. ([#124](https://github.com/argilla-io/argilla-server/pull/124))
+- Added `ARGILLA_SHOW_HUGGINGFACE_SPACE_PERSISTENT_STORAGE_WARNING` new environment variable to disable warning message when Hugging Face Spaces persistent storage is disabled. ([#124](https://github.com/argilla-io/argilla-server/pull/124))
 - Added `options_order` new settings attribute to support specify an order for options in multi label selection questions. ([#133](https://github.com/argilla-io/argilla-server/pull/133))
 - Added `POST /api/v1/datasets/:dataset_id/records/bulk` endpoint. ([#106](https://github.com/argilla-io/argilla-server/pull/106))
 - Added `PUT /api/v1/datasets/:dataset_id/records/bulk` endpoint. ([#106](https://github.com/argilla-io/argilla-server/pull/106))

src/argilla_server/contexts/settings.py

@@ -31,8 +31,8 @@ def _get_argilla_settings() -> ArgillaSettings:
     argilla_settings = ArgillaSettings()
 
     if _get_huggingface_settings():
-        argilla_settings.show_huggingface_space_persistant_storage_warning = (
-            settings.show_huggingface_space_persistant_storage_warning
+        argilla_settings.show_huggingface_space_persistent_storage_warning = (
+            settings.show_huggingface_space_persistent_storage_warning
         )
 
     return argilla_settings

src/argilla_server/schemas/v1/settings.py

@@ -18,7 +18,7 @@
 
 
 class ArgillaSettings(BaseModel):
-    show_huggingface_space_persistant_storage_warning: Optional[bool]
+    show_huggingface_space_persistent_storage_warning: Optional[bool]
 
 
 class HuggingfaceSettings(BaseSettings):
@@ -28,7 +28,9 @@ class HuggingfaceSettings(BaseSettings):
     space_host: str = Field(None, env="SPACE_HOST")
     space_repo_name: str = Field(None, env="SPACE_REPO_NAME")
     space_author_name: str = Field(None, env="SPACE_AUTHOR_NAME")
-    space_persistant_storage_enabled: bool = Field(False, env="PERSISTANT_STORAGE_ENABLED")
+    # NOTE: Hugging Face has a typo in their environment variable name,
+    # using PERSISTANT instead of PERSISTENT. We will use the correct spelling in our code.
+    space_persistent_storage_enabled: bool = Field(False, env="PERSISTANT_STORAGE_ENABLED")
 
     @property
     def is_running_on_huggingface(self) -> bool:

src/argilla_server/settings.py

@@ -128,9 +128,9 @@ class Settings(BaseSettings):
     )
 
     # Hugging Face settings
-    show_huggingface_space_persistant_storage_warning: bool = Field(
+    show_huggingface_space_persistent_storage_warning: bool = Field(
         default=True,
-        description="If True, show a warning when Hugging Face space persistant storage is disabled",
+        description="If True, show a warning when Hugging Face space persistent storage is disabled",
     )
 
     # See also the telemetry.py module

src/argilla_server/telemetry.py

@@ -25,7 +25,11 @@
 from argilla_server.constants import DEFAULT_USERNAME
 from argilla_server.models import User
 from argilla_server.settings import settings
-from argilla_server.utils._telemetry import is_running_on_docker_container, server_deployment_type
+from argilla_server.utils._telemetry import (
+    is_running_on_docker_container,
+    is_running_on_huggingface_space,
+    server_deployment_type,
+)
 
 try:
     from analytics import Client  # This import works only for version 2.2.0
@@ -54,13 +58,6 @@ def server_id(self) -> uuid.UUID:
     def __post_init__(self, enable_telemetry: bool, disable_send: bool, api_key: str, host: str):
         from argilla_server._version import __version__
 
-        self.client = None
-        if enable_telemetry:
-            try:
-                self.client = Client(write_key=api_key, gzip=True, host=host, send=not disable_send, max_retries=10)
-            except Exception as err:
-                _LOGGER.warning(f"Cannot initialize telemetry. Error: {err}. Disabling...")
-
         self._server_id = uuid.UUID(int=uuid.getnode())
         self._system_info = {
             "system": platform.system(),
@@ -74,19 +71,30 @@ def __post_init__(self, enable_telemetry: bool, disable_send: bool, api_key: str
         }
 
         _LOGGER.info("System Info:")
-        _LOGGER.info(json.dumps(self._system_info, indent=2))
+        _LOGGER.info(f"Server id: {self.server_id}")
+        _LOGGER.info(f"Context: {json.dumps(self._system_info, indent=2)}")
+
+        self.client: Optional[Client] = None
+        if enable_telemetry:
+            try:
+                client = Client(write_key=api_key, gzip=True, host=host, send=not disable_send, max_retries=10)
+                client.identify(user_id=str(self._server_id), traits=self._system_info)
+
+                self.client = client
+            except Exception as err:
+                _LOGGER.warning(f"Cannot initialize telemetry. Error: {err}. Disabling...")
 
     def track_data(self, action: str, data: Dict[str, Any], include_system_info: bool = True):
         if not self.client:
             return
 
         event_data = data.copy()
-        self.client.track(
-            user_id=str(self._server_id),
-            event=action,
-            properties=event_data,
-            context=self._system_info if include_system_info else {},
-        )
+
+        context = {}
+        if include_system_info:
+            context = self._system_info.copy()
+
+        self.client.track(user_id=str(self._server_id), event=action, properties=event_data, context=context)
 
 
 _CLIENT = TelemetryClient()

src/argilla_server/utils/_telemetry.py

@@ -11,39 +11,51 @@
 #  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
-
+import logging
 import os
 
-import psutil
+_LOGGER = logging.getLogger(__name__)
 
 
 def server_deployment_type() -> str:
     """Returns the type of deployment of the server."""
     global _server_deployment_type
 
     if _server_deployment_type is None:
-        _server_deployment_type = "server"
-
-        if is_running_on_docker_container() and _is_quickstart_env():
+        if is_running_on_huggingface_space():
+            _server_deployment_type = "huggingface_space"
+        elif _is_quickstart_env():
             _server_deployment_type = "quickstart"
-
+        else:
+            _server_deployment_type = "server"
     return _server_deployment_type
 
 
+def is_running_on_huggingface_space() -> bool:
+    """Returns True if the current process is running inside a Huggingface Space, False otherwise."""
+    from argilla_server.schemas.v1.settings import HuggingfaceSettings
+
+    return HuggingfaceSettings().is_running_on_huggingface
+
+
 def is_running_on_docker_container() -> bool:
     """Returns True if the current process is running inside a Docker container, False otherwise."""
     global _in_docker_container
 
     if _in_docker_container is None:
-        _in_docker_container = _has_docker_env() or _has_docker_cgroup()
+        if is_running_on_huggingface_space():
+            _in_docker_container = True
+        else:
+            _in_docker_container = _has_docker_env() or _has_docker_cgroup()
 
     return _in_docker_container
 
 
 def _has_docker_env() -> bool:
     try:
         return os.path.exists("/.dockerenv")
-    except:
+    except Exception as e:
+        _LOGGER.warning(f"Error while checking if running in Docker: {e}")
         return False
 
 
@@ -55,23 +67,11 @@ def _has_docker_cgroup() -> bool:
             and os.path.isfile(cgroup_path)
             and any("docker" in line for line in open(cgroup_path))
         )
-    except:
+    except Exception as e:
+        _LOGGER.warning(f"Error while checking if running in Docker: {e}")
         return False
 
 
-def _is_quickstart_script_running() -> bool:
-    # TODO: Any modification in the `quickstart.Dockerfile` file should be reflected here
-
-    try:
-        cmdline_key = "cmdline"
-        for process in psutil.process_iter([cmdline_key]):
-            if "start_quickstart_argilla.sh" in process.info[cmdline_key]:
-                return True
-    except:
-        pass
-    return False
-
-
 def _is_quickstart_env():
     # TODO: Any modification in the `quickstart.Dockerfile` file should be reflected here
 
@@ -84,7 +84,6 @@ def _is_quickstart_env():
         "ADMIN_API_KEY",
         "ANNOTATOR_USERNAME",
         "ANNOTATOR_PASSWORD",
-        "LOAD_DATASETS",
     ]:
         if env_var not in os.environ:
             return False

tests/unit/api/v1/settings/test_get_settings.py

@@ -34,26 +34,26 @@ async def test_get_settings_for_argilla_settings_running_on_huggingface(self, as
 
             assert response.status_code == 200
             assert response.json()["argilla"] == {
-                "show_huggingface_space_persistant_storage_warning": True,
+                "show_huggingface_space_persistent_storage_warning": True,
             }
 
     async def test_get_settings_for_argilla_settings_running_on_huggingface_with_disabled_storage_warning(
         self, async_client: AsyncClient
     ):
         with mock.patch.object(HUGGINGFACE_SETTINGS, "space_id", "space-id"):
-            with mock.patch.object(argilla_server_settings, "show_huggingface_space_persistant_storage_warning", False):
+            with mock.patch.object(argilla_server_settings, "show_huggingface_space_persistent_storage_warning", False):
                 response = await async_client.get(self.url())
 
                 assert response.status_code == 200
                 assert response.json()["argilla"] == {
-                    "show_huggingface_space_persistant_storage_warning": False,
+                    "show_huggingface_space_persistent_storage_warning": False,
                 }
 
     async def test_get_settings_for_argilla_settings_not_running_on_huggingface(self, async_client: AsyncClient):
         response = await async_client.get(self.url())
 
         assert response.status_code == 200
-        assert "show_huggingface_space_persistant_storage_warning" not in response.json()["argilla"]
+        assert "show_huggingface_space_persistent_storage_warning" not in response.json()["argilla"]
 
     async def test_get_settings_for_huggingface_settings_running_on_huggingface(self, async_client: AsyncClient):
         huggingface_os_environ = {
@@ -78,7 +78,7 @@ async def test_get_settings_for_huggingface_settings_running_on_huggingface(self
                     "space_host": "space-host",
                     "space_repo_name": "space-repo-name",
                     "space_author_name": "space-author-name",
-                    "space_persistant_storage_enabled": True,
+                    "space_persistent_storage_enabled": True,
                 }
 
     async def test_get_settings_for_huggingface_settings_not_running_on_huggingface(self, async_client: AsyncClient):