feat: change PipelineConfigError to DocumentStoreError with more deta…

…ils (deepset-ai#3783)

haystack/pipelines/base.py

@@ -46,7 +46,7 @@
 from haystack.pipelines.utils import generate_code, print_eval_report
 from haystack.utils import DeepsetCloud, calculate_context_similarity
 from haystack.schema import Answer, EvaluationResult, MultiLabel, Document, Span
-from haystack.errors import HaystackError, PipelineError, PipelineConfigError
+from haystack.errors import HaystackError, PipelineError, PipelineConfigError, DocumentStoreError
 from haystack.nodes import BaseGenerator, Docs2Answers, BaseReader, BaseSummarizer, BaseTranslator, QuestionGenerator
 from haystack.nodes.base import BaseComponent, RootNode
 from haystack.nodes.retriever.base import BaseRetriever
@@ -2022,9 +2022,13 @@ def _load_or_get_component(cls, name: str, definitions: dict, components: dict):
                 f"Failed loading pipeline component '{name}': "
                 "seems like the component does not exist. Did you spell its name correctly?"
             ) from ke
+        except ConnectionError as ce:
+            raise DocumentStoreError(f"Failed loading pipeline component '{name}': '{ce}'") from ce
+        except DocumentStoreError as de:
+            raise de
         except Exception as e:
             raise PipelineConfigError(
-                f"Failed loading pipeline component '{name}'. " "See the stacktrace above for more informations."
+                f"Failed loading pipeline component '{name}'. See the stacktrace above for more information."
             ) from e
 
     def save_to_yaml(self, path: Path, return_defaults: bool = False):
@@ -2152,7 +2156,7 @@ def _validate_node_names_in_params(self, params: Optional[Dict]):
                 not_a_node = set(params.keys()) - set(self.graph.nodes)
                 # "debug" will be picked up by _dispatch_run, see its code
                 # "add_isolated_node_eval" is set by pipeline.eval / pipeline.eval_batch
-                valid_global_params = set(["debug", "add_isolated_node_eval"])
+                valid_global_params = {"debug", "add_isolated_node_eval"}
                 for node_id in self.graph.nodes:
                     run_signature_args = self._get_run_node_signature(node_id)
                     valid_global_params |= set(run_signature_args)

test/pipelines/test_pipeline_yaml.py

@@ -14,7 +14,7 @@
 from haystack import Pipeline
 from haystack.nodes import _json_schema
 from haystack.nodes import FileTypeClassifier
-from haystack.errors import HaystackError, PipelineConfigError, PipelineSchemaError
+from haystack.errors import HaystackError, PipelineConfigError, PipelineSchemaError, DocumentStoreError
 from haystack.nodes.base import BaseComponent
 
 from ..conftest import SAMPLES_PATH, MockNode, MockDocumentStore, MockReader, MockRetriever
@@ -141,6 +141,46 @@ def test_load_yaml(tmp_path):
     assert isinstance(pipeline.get_node("reader"), MockReader)
 
 
+def test_load_yaml_elasticsearch_not_responding(tmp_path):
+    # Test if DocumentStoreError is raised if elasticsearch instance is not responding (due to wrong port)
+    with open(tmp_path / "tmp_config.yml", "w") as tmp_file:
+        tmp_file.write(
+            f"""
+            version: ignore
+            components:
+            - name: ESRetriever
+              type: BM25Retriever
+              params:
+                document_store: DocumentStore
+            - name: DocumentStore
+              type: ElasticsearchDocumentStore
+              params:
+                port: 1234
+            - name: PDFConverter
+              type: PDFToTextConverter
+            - name: Preprocessor
+              type: PreProcessor
+            pipelines:
+            - name: query_pipeline
+              nodes:
+              - name: ESRetriever
+                inputs: [Query]
+            - name: indexing_pipeline
+              nodes:
+              - name: PDFConverter
+                inputs: [File]
+              - name: Preprocessor
+                inputs: [PDFConverter]
+              - name: ESRetriever
+                inputs: [Preprocessor]
+              - name: DocumentStore
+                inputs: [ESRetriever]
+        """
+        )
+    with pytest.raises(DocumentStoreError):
+        Pipeline.load_from_yaml(path=tmp_path / "tmp_config.yml", pipeline_name="indexing_pipeline")
+
+
 def test_load_yaml_non_existing_file():
     with pytest.raises(FileNotFoundError):
         Pipeline.load_from_yaml(path=SAMPLES_PATH / "pipeline" / "I_dont_exist.yml")