(service) default results_dir and lancedb_uri to the current working …

nemo_retriever/src/nemo_retriever/service/cli.py

@@ -61,6 +61,16 @@ def start(
         None, "--gpu-devices", help="Comma-separated GPU device IDs (overrides YAML)."
     ),
     db_path: Optional[str] = typer.Option(None, "--db-path", help="SQLite database path (overrides YAML)."),
+    results_dir: Optional[str] = typer.Option(
+        None,
+        "--results-dir",
+        help="Directory for per-job result JSON files (overrides YAML).",
+    ),
+    lancedb_uri: Optional[str] = typer.Option(
+        None,
+        "--lancedb-uri",
+        help="LanceDB URI for the embeddings table (overrides YAML).",
+    ),
     api_token: Optional[str] = typer.Option(
         None,
         "--api-token",
@@ -108,6 +118,10 @@ def start(
         overrides["resources.gpu_devices"] = [d.strip() for d in gpu_devices.split(",") if d.strip()]
     if db_path is not None:
         overrides["database.path"] = db_path
+    if results_dir is not None:
+        overrides["processing.results_dir"] = results_dir
+    if lancedb_uri is not None:
+        overrides["vector_store.lancedb_uri"] = lancedb_uri
     if api_token is not None:
         overrides["auth.api_token"] = api_token
     if drain_timeout_s is not None:

nemo_retriever/src/nemo_retriever/service/config.py

@@ -55,6 +55,7 @@ class ProcessingConfig(BaseModel):
     num_workers: int = 16
     batch_size: int = 32
     batch_timeout_s: float = 2.0
+    # Relative paths resolve against the current working directory.
     results_dir: str = "retriever_results"
 
 
@@ -128,7 +129,8 @@ class VectorStoreConfig(BaseModel):
 
     model_config = ConfigDict(extra="forbid")
 
-    lancedb_uri: str = "/var/lib/nemo-retriever/lancedb"
+    # Relative paths resolve against the current working directory.
+    lancedb_uri: str = "lancedb"
     lancedb_table: str = "nv-ingest"
     top_k: int = 10
     vector_column_name: str = "vector"

nemo_retriever/src/nemo_retriever/service/retriever-service.yaml

@@ -37,7 +37,11 @@ processing:
   # Directory where per-job page results are written as JSON files.
   # Each job gets a subdirectory: {results_dir}/{job_id}/.  The results
   # endpoint reads from here instead of SQLite for faster retrieval.
-  results_dir: "/home/local/jdyer/datasets/retriever_results"
+  #
+  # Relative paths resolve against the current working directory so
+  # `retriever service start` works out of the box on any host.  The
+  # Helm chart overrides this to a path on the persistent volume.
+  results_dir: "retriever_results"
 
 # Remote NIM microservice endpoints.  When set, the pipeline calls these
 # HTTP endpoints instead of loading GPU models locally.  All four must
@@ -105,10 +109,11 @@ spool:
 # to this table after every batch; the /v1/query endpoint reads from it.
 # Both sides MUST point at the same URI + table for query to work.
 #
-# Override lancedb_uri for local development — the container default
-# (/var/lib/nemo-retriever/lancedb) won't exist on the host.
+# Relative paths resolve against the current working directory.
+# Override via ./retriever-service.yaml or `--lancedb-uri` for any
+# deployment that should keep its embeddings on a different volume.
 vector_store:
-  lancedb_uri: "/home/local/jdyer/Development/NeMo-Retriever/lancedb"
+  lancedb_uri: "lancedb"
   lancedb_table: "nv-ingest"
   top_k: 10
   vector_column_name: "vector"

nemo_retriever/tests/test_service_bundled_config.py

@@ -0,0 +1,35 @@
+# SPDX-FileCopyrightText: Copyright (c) 2026, NVIDIA CORPORATION & AFFILIATES.
+# All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+
+"""Tests for the bundled ``retriever-service.yaml`` shipped with the package."""
+
+from __future__ import annotations
+
+from pathlib import Path
+
+from nemo_retriever.service.config import _bundled_yaml_path, load_config
+
+
+def test_bundled_yaml_loads_cleanly():
+    bundled = _bundled_yaml_path()
+    assert bundled.is_file()
+    cfg = load_config(config_path=str(bundled))
+    assert cfg.processing.num_workers > 0
+    assert cfg.processing.results_dir
+    assert cfg.vector_store.lancedb_uri
+
+
+def test_bundled_yaml_uses_cwd_relative_paths(tmp_path, monkeypatch):
+    """`retriever service start` with no overrides must work in any cwd —
+    the bundled YAML's writable-state defaults must be relative.
+    """
+    monkeypatch.chdir(tmp_path)  # avoid picking up a stray ./retriever-service.yaml
+    cfg = load_config(config_path=None)
+
+    assert not Path(
+        cfg.processing.results_dir
+    ).is_absolute(), f"bundled YAML pins absolute results_dir: {cfg.processing.results_dir!r}"
+    assert not Path(
+        cfg.vector_store.lancedb_uri
+    ).is_absolute(), f"bundled YAML pins absolute lancedb_uri: {cfg.vector_store.lancedb_uri!r}"