acryl-datahub 0.15.0rc1__py3-none-any.whl → 0.15.0rc3__py3-none-any.whl

{acryl_datahub-0.15.0rc1.dist-info → acryl_datahub-0.15.0rc3.dist-info}/RECORD

@@ -1,4 +1,4 @@
-datahub/__init__.py,sha256=MoEVuYLqJQKtHGXciQ4YfjAktrZuIswXVPeaXMasj-8,574
+datahub/__init__.py,sha256=bjNaEX_th93Zf5oFOCSAaZbNy-DUb3y4CxlHrAvZOkI,574
 datahub/__main__.py,sha256=pegIvQ9hzK7IhqVeUi1MeADSZ2QlP-D3K0OQdEg55RU,106
 datahub/entrypoints.py,sha256=3-qSfXAx3Z0FEkBV5tlO8fQr4xk4ySeDRMVTpS5Xd6A,7793
 datahub/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -162,7 +162,7 @@ datahub/ingestion/glossary/classifier.py,sha256=daLxnVv_JlfB_jBOxH5LrU_xQRndrsGo
 datahub/ingestion/glossary/classifier_registry.py,sha256=yFOYLQhDgCLqXYMG3L1BquXafeLcZDcmp8meyw6k9ts,307
 datahub/ingestion/glossary/datahub_classifier.py,sha256=8VhwuLDhyOqqOr0jqAPIgorb4eAOnvTr4m13Y2Wy1-E,7515
 datahub/ingestion/graph/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-datahub/ingestion/graph/client.py,sha256=-SpQq0zWJ9hoeG9YhWUVZgPB97DD78AsgFJgpOsdAZ0,64476
+datahub/ingestion/graph/client.py,sha256=oBlM6RSo0SPFJ-yit2eFFOB3rOpnjKtQ83YNiWGd334,64584
 datahub/ingestion/graph/config.py,sha256=3b_Gxa5wcBnphP63bBiAFdWS7PJhUHRE1WZL_q4Cw8k,749
 datahub/ingestion/graph/connections.py,sha256=9462L0ZWGKURyypAln25eMPhK3pcufBar9tNDoqspXs,741
 datahub/ingestion/graph/filters.py,sha256=UeUZQHoimavIYx-jXLA0WGkOUe10TaO8uEZkfa-QgNE,6188
@@ -172,7 +172,7 @@ datahub/ingestion/reporting/file_reporter.py,sha256=tiWukmMxHrTQI3rOAumsq6lRlw8T
 datahub/ingestion/reporting/reporting_provider_registry.py,sha256=jTYSh3T4sensjnHQfPLiIcbA2dG8w0px9ghChAJjGdU,310
 datahub/ingestion/run/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/ingestion/run/connection.py,sha256=dqS9Fp8byIJNydPmVgtjjjlPJguuUWuMuvGnpNbQdSs,1474
-datahub/ingestion/run/pipeline.py,sha256=QV1i1TWCIH9gBDGe8Xs0JEbOqEUmWbhUhfx7gvrR7vc,30548
+datahub/ingestion/run/pipeline.py,sha256=8MNUC19h7AvxjlDJj3E_FZlY56SAUlYG0heIko2XK_g,30572
 datahub/ingestion/run/pipeline_config.py,sha256=91Uvs76EGbCzZZbm819TT0L6pixf2tfI2_nHpnCoyS4,3948
 datahub/ingestion/sink/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/ingestion/sink/blackhole.py,sha256=-jYcWo4i8q7312bCIoHrGr7nT9JdPvA7c4jvSc7YOgY,557
@@ -300,8 +300,8 @@ datahub/ingestion/source/fivetran/fivetran.py,sha256=uKbM5czPz-6LOseoh1FwavWDIuL
 datahub/ingestion/source/fivetran/fivetran_log_api.py,sha256=EAak3hJpe75WZSgz6wP_CyAT5Cian2N4a-lb8x1NKHk,12776
 datahub/ingestion/source/fivetran/fivetran_query.py,sha256=vLrTj7e-0NxZ2U4bWTB57pih42WirqPlUvwtIRfStlQ,5275
 datahub/ingestion/source/gc/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-datahub/ingestion/source/gc/datahub_gc.py,sha256=p1LiiZJDMaEjWuhnT5t83ALWDEHcPqmoZX64fCBGYmQ,11645
-datahub/ingestion/source/gc/dataprocess_cleanup.py,sha256=Y4KluNVVSdUbE85jPba8oc_EKm8WmKJrIbAuTPnSzx0,14301
+datahub/ingestion/source/gc/datahub_gc.py,sha256=f6Erj3KfD0Hx3ydwL5MUVCZgFzS9c6U2Pkr54JLIUOA,12394
+datahub/ingestion/source/gc/dataprocess_cleanup.py,sha256=IEEHO6UvDWWK3W5siqFrk4J1zUKbL6TrKNUaXdNiEW4,14362
 datahub/ingestion/source/gc/execution_request_cleanup.py,sha256=cHJmxz4NmA7VjTX2iGEo3wZ_SDrjC_rCQcnRxKgfUVI,8713
 datahub/ingestion/source/gc/soft_deleted_entity_cleanup.py,sha256=qLgdr-Rrsba0z_Y-CaHT9d1zSgy2jzg6CXaCKoN2jFk,7360
 datahub/ingestion/source/gcs/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -872,7 +872,7 @@ datahub/sql_parsing/sqlglot_utils.py,sha256=8MYzkyekhup3ihVStRPuwneWPNu17xhBg5SG
 datahub/sql_parsing/tool_meta_extractor.py,sha256=pE-pkRKBfNTXEJkaQM9NlG807mc-X6OtetgskJySCs8,2908
 datahub/telemetry/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/telemetry/stats.py,sha256=YltbtC3fe6rl1kcxn1A-mSnVpECTPm5k-brrUt7QxTI,967
-datahub/telemetry/telemetry.py,sha256=xXtvMVkX1YNu4Z0NUZnDAiIYP6c9mFujbGpd1wk3bgM,14763
+datahub/telemetry/telemetry.py,sha256=gzla-QGNsynGg2FqFxiDDFQ0emG53MJ9lhOA2-UUg-Y,15047
 datahub/testing/__init__.py,sha256=TywIuzGQvzJsNhI_PGD1RFk11M3RtGl9jIMtAVVHIkg,272
 datahub/testing/check_imports.py,sha256=EKuJmgUA46uOrlaOy0fCvPB7j9POkpJ0ExhO_pT3YAk,1356
 datahub/testing/check_sql_parser_result.py,sha256=f7U7IUSbfV4VACdNI857wPZ9tAZ9j6mXiXmcJNT_RzM,2671
@@ -971,8 +971,8 @@ datahub_provider/operators/datahub_assertion_operator.py,sha256=uvTQ-jk2F0sbqqxp
 datahub_provider/operators/datahub_assertion_sensor.py,sha256=lCBj_3x1cf5GMNpHdfkpHuyHfVxsm6ff5x2Z5iizcAo,140
 datahub_provider/operators/datahub_operation_operator.py,sha256=aevDp2FzX7FxGlXrR0khoHNbxbhKR2qPEX5e8O2Jyzw,174
 datahub_provider/operators/datahub_operation_sensor.py,sha256=8fcdVBCEPgqy1etTXgLoiHoJrRt_nzFZQMdSzHqSG7M,168
-acryl_datahub-0.15.0rc1.dist-info/METADATA,sha256=bar0qt0g6b-U-fWNB2iUu0HtuFOWGNhm_8lrnyK7iSg,171129
-acryl_datahub-0.15.0rc1.dist-info/WHEEL,sha256=PZUExdf71Ui_so67QXpySuHtCi3-J3wvF4ORK6k_S8U,91
-acryl_datahub-0.15.0rc1.dist-info/entry_points.txt,sha256=VcQx0dnqaYLyeY_L5OaX7bLmmE-Il7TAXkxCKvEn2bA,9432
-acryl_datahub-0.15.0rc1.dist-info/top_level.txt,sha256=iLjSrLK5ox1YVYcglRUkcvfZPvKlobBWx7CTUXx8_GI,25
-acryl_datahub-0.15.0rc1.dist-info/RECORD,,
+acryl_datahub-0.15.0rc3.dist-info/METADATA,sha256=l71CWZc9fD8L_4p0NFRk9ruU6w4QEZgGJq3RObW_ljA,171129
+acryl_datahub-0.15.0rc3.dist-info/WHEEL,sha256=PZUExdf71Ui_so67QXpySuHtCi3-J3wvF4ORK6k_S8U,91
+acryl_datahub-0.15.0rc3.dist-info/entry_points.txt,sha256=VcQx0dnqaYLyeY_L5OaX7bLmmE-Il7TAXkxCKvEn2bA,9432
+acryl_datahub-0.15.0rc3.dist-info/top_level.txt,sha256=iLjSrLK5ox1YVYcglRUkcvfZPvKlobBWx7CTUXx8_GI,25
+acryl_datahub-0.15.0rc3.dist-info/RECORD,,

datahub/__init__.py

@@ -3,7 +3,7 @@ import warnings
 
 # Published at https://pypi.org/project/acryl-datahub/.
 __package_name__ = "acryl-datahub"
-__version__ = "0.15.0rc1"
+__version__ = "0.15.0rc3"
 
 
 def is_dev_mode() -> bool:

datahub/ingestion/graph/client.py

@@ -67,6 +67,7 @@ from datahub.metadata.schema_classes import (
     SystemMetadataClass,
     TelemetryClientIdClass,
 )
+from datahub.telemetry.telemetry import telemetry_instance
 from datahub.utilities.perf_timer import PerfTimer
 from datahub.utilities.str_enum import StrEnum
 from datahub.utilities.urns.urn import Urn, guess_entity_type
@@ -1819,4 +1820,5 @@ def get_default_graph() -> DataHubGraph:
     graph_config = config_utils.load_client_config()
     graph = DataHubGraph(graph_config)
     graph.test_connection()
+    telemetry_instance.set_context(server=graph)
     return graph

datahub/ingestion/run/pipeline.py

@@ -44,7 +44,8 @@ from datahub.ingestion.transformer.system_metadata_transformer import (
 )
 from datahub.ingestion.transformer.transform_registry import transform_registry
 from datahub.metadata.schema_classes import MetadataChangeProposalClass
-from datahub.telemetry import stats, telemetry
+from datahub.telemetry import stats
+from datahub.telemetry.telemetry import telemetry_instance
 from datahub.utilities._custom_package_loader import model_version_name
 from datahub.utilities.global_warning_util import (
     clear_global_warnings,
@@ -273,8 +274,9 @@ class Pipeline:
         if self.graph is None and isinstance(self.sink, DatahubRestSink):
             with _add_init_error_context("setup default datahub client"):
                 self.graph = self.sink.emitter.to_graph()
+                self.graph.test_connection()
         self.ctx.graph = self.graph
-        telemetry.telemetry_instance.update_capture_exception_context(server=self.graph)
+        telemetry_instance.set_context(server=self.graph)
 
         with set_graph_context(self.graph):
             with _add_init_error_context("configure reporters"):
@@ -615,7 +617,7 @@ class Pipeline:
         sink_warnings = len(self.sink.get_report().warnings)
         global_warnings = len(get_global_warnings())
 
-        telemetry.telemetry_instance.ping(
+        telemetry_instance.ping(
             "ingest_stats",
             {
                 "source_type": self.source_type,
@@ -637,7 +639,6 @@ class Pipeline:
                 ),
                 "has_pipeline_name": bool(self.config.pipeline_name),
             },
-            self.ctx.graph,
         )
 
     def _approx_all_vals(self, d: LossyList[Any]) -> int:

datahub/ingestion/source/gc/datahub_gc.py

@@ -144,15 +144,32 @@ class DataHubGcSource(Source):
         self,
     ) -> Iterable[MetadataWorkUnit]:
         if self.config.cleanup_expired_tokens:
-            self.revoke_expired_tokens()
+            try:
+                self.revoke_expired_tokens()
+            except Exception as e:
+                self.report.failure("While trying to cleanup expired token ", exc=e)
         if self.config.truncate_indices:
-            self.truncate_indices()
+            try:
+                self.truncate_indices()
+            except Exception as e:
+                self.report.failure("While trying to truncate indices ", exc=e)
         if self.dataprocess_cleanup:
-            yield from self.dataprocess_cleanup.get_workunits_internal()
+            try:
+                yield from self.dataprocess_cleanup.get_workunits_internal()
+            except Exception as e:
+                self.report.failure("While trying to cleanup data process ", exc=e)
         if self.soft_deleted_entities_cleanup:
-            self.soft_deleted_entities_cleanup.cleanup_soft_deleted_entities()
+            try:
+                self.soft_deleted_entities_cleanup.cleanup_soft_deleted_entities()
+            except Exception as e:
+                self.report.failure(
+                    "While trying to cleanup soft deleted entities ", exc=e
+                )
         if self.execution_request_cleanup:
-            self.execution_request_cleanup.run()
+            try:
+                self.execution_request_cleanup.run()
+            except Exception as e:
+                self.report.failure("While trying to cleanup execution request ", exc=e)
         yield from []
 
     def truncate_indices(self) -> None:

datahub/ingestion/source/gc/dataprocess_cleanup.py

@@ -404,7 +404,9 @@ class DataProcessCleanup:
                     try:
                         self.delete_dpi_from_datajobs(datajob_entity)
                     except Exception as e:
-                        logger.error(f"While trying to delete {datajob_entity} got {e}")
+                        self.report.failure(
+                            f"While trying to delete {datajob_entity} ", exc=e
+                        )
                 if (
                     datajob_entity.total_runs == 0
                     and self.config.delete_empty_data_jobs

datahub/telemetry/telemetry.py

@@ -7,7 +7,7 @@ import sys
 import uuid
 from functools import wraps
 from pathlib import Path
-from typing import Any, Callable, Dict, List, Optional, TypeVar
+from typing import TYPE_CHECKING, Any, Callable, Dict, List, Optional, TypeVar
 
 from mixpanel import Consumer, Mixpanel
 from typing_extensions import ParamSpec
@@ -16,10 +16,12 @@ import datahub as datahub_package
 from datahub.cli.config_utils import DATAHUB_ROOT_FOLDER
 from datahub.cli.env_utils import get_boolean_env_variable
 from datahub.configuration.common import ExceptionWithProps
-from datahub.ingestion.graph.client import DataHubGraph
 from datahub.metadata.schema_classes import _custom_package_path
 from datahub.utilities.perf_timer import PerfTimer
 
+if TYPE_CHECKING:
+    from datahub.ingestion.graph.client import DataHubGraph
+
 logger = logging.getLogger(__name__)
 
 DATAHUB_FOLDER = Path(DATAHUB_ROOT_FOLDER)
@@ -117,7 +119,11 @@ class Telemetry:
     tracking_init: bool = False
     sentry_enabled: bool = False
 
+    context_properties: Dict[str, Any] = {}
+
     def __init__(self):
+        self.context_properties = {}
+
         if SENTRY_DSN:
             self.sentry_enabled = True
             try:
@@ -157,6 +163,9 @@ class Telemetry:
             except Exception as e:
                 logger.debug(f"Error connecting to mixpanel: {e}")
 
+        # Initialize the default properties for all events.
+        self.set_context()
+
     def update_config(self) -> bool:
         """
         Update the config file with the current client ID and enabled status.
@@ -238,18 +247,22 @@ class Telemetry:
 
         return False
 
-    def update_capture_exception_context(
+    def set_context(
         self,
-        server: Optional[DataHubGraph] = None,
+        server: Optional["DataHubGraph"] = None,
         properties: Optional[Dict[str, Any]] = None,
     ) -> None:
+        self.context_properties = {
+            **self._server_props(server),
+            **(properties or {}),
+        }
+
         if self.sentry_enabled:
             from sentry_sdk import set_tag
 
             properties = {
                 **_default_telemetry_properties(),
-                **self._server_props(server),
-                **(properties or {}),
+                **self.context_properties,
             }
 
             for key in properties:
@@ -297,7 +310,6 @@ class Telemetry:
         self,
         event_name: str,
         properties: Optional[Dict[str, Any]] = None,
-        server: Optional[DataHubGraph] = None,
     ) -> None:
         """
         Send a single telemetry event.
@@ -323,14 +335,15 @@ class Telemetry:
 
             properties = {
                 **_default_telemetry_properties(),
-                **self._server_props(server),
+                **self.context_properties,
                 **properties,
             }
             self.mp.track(self.client_id, event_name, properties)
         except Exception as e:
             logger.debug(f"Error reporting telemetry: {e}")
 
-    def _server_props(self, server: Optional[DataHubGraph]) -> Dict[str, str]:
+    @classmethod
+    def _server_props(cls, server: Optional["DataHubGraph"]) -> Dict[str, str]:
         if not server:
             return {
                 "server_type": "n/a",
@@ -435,6 +448,7 @@ def with_telemetry(
                             **call_props,
                             "status": "error",
                             **_error_props(e),
+                            "code": e.code,
                         },
                     )
                 telemetry_instance.capture_exception(e)