feat(ingest/looker): cleanup error handling (#9135)

datahub-project · Oct 30, 2023 · 51d6d1f · 51d6d1f
1 parent 58bcedc
commit 51d6d1f
Show file tree

Hide file tree

Showing 4 changed files with 14 additions and 19 deletions.
diff --git a/metadata-ingestion/src/datahub/ingestion/api/workunit.py b/metadata-ingestion/src/datahub/ingestion/api/workunit.py
@@ -22,7 +22,11 @@ class MetadataWorkUnit(WorkUnit):
     metadata: Union[
         MetadataChangeEvent, MetadataChangeProposal, MetadataChangeProposalWrapper
     ]
-    # A workunit creator can determine if this workunit is allowed to fail
+
+    # A workunit creator can determine if this workunit is allowed to fail.
+    # TODO: This flag was initially added during the rollout of the subType aspect
+    # to improve backwards compatibility, but is not really needed anymore and so
+    # should be removed.
     treat_errors_as_warnings: bool = False
 
     # When this is set to false, this MWU will be ignored by automatic helpers

diff --git a/metadata-ingestion/src/datahub/ingestion/source/looker/looker_lib_wrapper.py b/metadata-ingestion/src/datahub/ingestion/source/looker/looker_lib_wrapper.py
@@ -123,8 +123,12 @@ def get_user(self, id_: str, user_fields: str) -> Optional[User]:
                 transport_options=self.transport_options,
             )
         except SDKError as e:
-            logger.warning(f"Could not find user with id {id_}")
-            logger.warning(f"Failure was {e}")
+            if "Looker Not Found (404)" in str(e):
+                # User not found
+                logger.info(f"Could not find user with id {id_}: 404 error")
+            else:
+                logger.warning(f"Could not find user with id {id_}")
+                logger.warning(f"Failure was {e}")
         # User not found
         return None
 

diff --git a/metadata-ingestion/src/datahub/ingestion/source/looker/looker_source.py b/metadata-ingestion/src/datahub/ingestion/source/looker/looker_source.py
@@ -926,14 +926,7 @@ def process_metrics_dimensions_and_fields_for_dashboard(
         mcps = chart_mcps
         mcps.append(dashboard_mcp)
 
-        workunits = [
-            MetadataWorkUnit(
-                id=f"looker-{mcp.aspectName}-{mcp.entityUrn}",
-                mcp=mcp,
-                treat_errors_as_warnings=True,
-            )
-            for mcp in mcps
-        ]
+        workunits = [mcp.as_workunit() for mcp in mcps]
 
         return workunits
 
@@ -1320,10 +1313,7 @@ def get_workunits_internal(self) -> Iterable[MetadataWorkUnit]:
                     id=f"looker-{event.proposedSnapshot.urn}", mce=event
                 )
             elif isinstance(event, MetadataChangeProposalWrapper):
-                # We want to treat subtype aspects as optional, so allowing failures in this aspect to be treated as warnings rather than failures
-                yield event.as_workunit(
-                    treat_errors_as_warnings=event.aspectName in ["subTypes"]
-                )
+                yield event.as_workunit()
             else:
                 raise Exception(f"Unexpected type of event {event}")
         self.reporter.report_stage_end("explore_metadata")

diff --git a/metadata-ingestion/src/datahub/ingestion/source/looker/lookml_source.py b/metadata-ingestion/src/datahub/ingestion/source/looker/lookml_source.py
@@ -2171,10 +2171,7 @@ def get_internal_workunits(self) -> Iterable[MetadataWorkUnit]:  # noqa: C901
                                     for mcp in self._build_dataset_mcps(
                                         maybe_looker_view
                                     ):
-                                        # We want to treat mcp aspects as optional, so allowing failures in this aspect to be treated as warnings rather than failures
-                                        yield mcp.as_workunit(
-                                            treat_errors_as_warnings=True
-                                        )
+                                        yield mcp.as_workunit()
                                 else:
                                     (
                                         prev_model_name,