nv-ingest 2025.8.2.dev20250802__tar.gz → 2025.8.4.dev20250804__tar.gz

{nv_ingest-2025.8.2.dev20250802 → nv_ingest-2025.8.4.dev20250804}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nv-ingest
-Version: 2025.8.2.dev20250802
+Version: 2025.8.4.dev20250804
 Summary: Python module for multimodal document ingestion
 Author-email: Jeremy Dyer <jdyer@nvidia.com>
 License:                                  Apache License

{nv_ingest-2025.8.2.dev20250802 → nv_ingest-2025.8.4.dev20250804}/nv_ingest/framework/orchestration/ray/util/pipeline/pipeline_builders.py

@@ -9,6 +9,7 @@ import os
 from typing import Dict, Any
 
 import ray
+from ray import LoggingConfig
 from pydantic import BaseModel
 
 from nv_ingest.framework.orchestration.ray.primitives.ray_pipeline import RayPipeline
@@ -47,16 +48,174 @@ def export_config_to_env(ingest_config: Any) -> None:
     os.environ.update({key.upper(): val for key, val in ingest_config.items()})
 
 
+def build_logging_config_from_env() -> LoggingConfig:
+    """
+    Build Ray LoggingConfig from environment variables.
+
+    Package-level preset (sets all defaults):
+    - INGEST_RAY_LOG_LEVEL: PRODUCTION, DEVELOPMENT, DEBUG. Default: DEVELOPMENT
+
+    Individual environment variables (override preset defaults):
+    - RAY_LOGGING_LEVEL: Log level (DEBUG, INFO, WARNING, ERROR, CRITICAL). Default: INFO
+    - RAY_LOGGING_ENCODING: Log encoding format (TEXT, JSON). Default: TEXT
+    - RAY_LOGGING_ADDITIONAL_ATTRS: Comma-separated list of additional standard logger attributes
+    - RAY_DEDUP_LOGS: Enable/disable log deduplication (0/1). Default: 1 (enabled)
+    - RAY_LOG_TO_DRIVER: Enable/disable logging to driver (true/false). Default: true
+    - RAY_LOGGING_ROTATE_BYTES: Maximum log file size before rotation (bytes). Default: 1GB
+    - RAY_LOGGING_ROTATE_BACKUP_COUNT: Number of backup log files to keep. Default: 19
+    - RAY_DISABLE_IMPORT_WARNING: Disable Ray import warnings (0/1). Default: 0
+    - RAY_USAGE_STATS_ENABLED: Enable/disable usage stats collection (0/1). Default: 1
+    """
+
+    # Apply package-level preset defaults first
+    preset_level = os.environ.get("INGEST_RAY_LOG_LEVEL", "DEVELOPMENT").upper()
+
+    # Define preset configurations
+    presets = {
+        "PRODUCTION": {
+            "RAY_LOGGING_LEVEL": "ERROR",
+            "RAY_LOGGING_ENCODING": "TEXT",
+            "RAY_LOGGING_ADDITIONAL_ATTRS": "",
+            "RAY_DEDUP_LOGS": "1",
+            "RAY_LOG_TO_DRIVER": "0",  # false
+            "RAY_LOGGING_ROTATE_BYTES": "1073741824",  # 1GB
+            "RAY_LOGGING_ROTATE_BACKUP_COUNT": "9",  # 10GB total
+            "RAY_DISABLE_IMPORT_WARNING": "1",
+            "RAY_USAGE_STATS_ENABLED": "0",
+        },
+        "DEVELOPMENT": {
+            "RAY_LOGGING_LEVEL": "INFO",
+            "RAY_LOGGING_ENCODING": "TEXT",
+            "RAY_LOGGING_ADDITIONAL_ATTRS": "",
+            "RAY_DEDUP_LOGS": "1",
+            "RAY_LOG_TO_DRIVER": "1",  # true
+            "RAY_LOGGING_ROTATE_BYTES": "1073741824",  # 1GB
+            "RAY_LOGGING_ROTATE_BACKUP_COUNT": "19",  # 20GB total
+            "RAY_DISABLE_IMPORT_WARNING": "0",
+            "RAY_USAGE_STATS_ENABLED": "1",
+        },
+        "DEBUG": {
+            "RAY_LOGGING_LEVEL": "DEBUG",
+            "RAY_LOGGING_ENCODING": "JSON",
+            "RAY_LOGGING_ADDITIONAL_ATTRS": "name,funcName,lineno",
+            "RAY_DEDUP_LOGS": "0",
+            "RAY_LOG_TO_DRIVER": "1",  # true
+            "RAY_LOGGING_ROTATE_BYTES": "536870912",  # 512MB
+            "RAY_LOGGING_ROTATE_BACKUP_COUNT": "39",  # 20GB total
+            "RAY_DISABLE_IMPORT_WARNING": "0",
+            "RAY_USAGE_STATS_ENABLED": "1",
+        },
+    }
+
+    # Validate preset level
+    if preset_level not in presets:
+        logger.warning(
+            f"Invalid INGEST_RAY_LOG_LEVEL '{preset_level}', using DEVELOPMENT. "
+            f"Valid presets: {list(presets.keys())}"
+        )
+        preset_level = "DEVELOPMENT"
+
+    # Apply preset defaults (only if env var not already set)
+    preset_config = presets[preset_level]
+    for key, default_value in preset_config.items():
+        if key not in os.environ:
+            os.environ[key] = default_value
+
+    logger.info(f"Applied Ray logging preset: {preset_level}")
+
+    # Get log level from environment, default to INFO
+    log_level = os.environ.get("RAY_LOGGING_LEVEL", "INFO").upper()
+
+    # Validate log level
+    valid_levels = ["DEBUG", "INFO", "WARNING", "ERROR", "CRITICAL"]
+    if log_level not in valid_levels:
+        logger.warning(f"Invalid RAY_LOGGING_LEVEL '{log_level}', using INFO. Valid levels: {valid_levels}")
+        log_level = "INFO"
+
+    # Get encoding format from environment, default to TEXT
+    encoding = os.environ.get("RAY_LOGGING_ENCODING", "TEXT").upper()
+
+    # Validate encoding
+    valid_encodings = ["TEXT", "JSON"]
+    if encoding not in valid_encodings:
+        logger.warning(f"Invalid RAY_LOGGING_ENCODING '{encoding}', using TEXT. Valid encodings: {valid_encodings}")
+        encoding = "TEXT"
+
+    # Get additional standard logger attributes
+    additional_attrs_str = os.environ.get("RAY_LOGGING_ADDITIONAL_ATTRS", "")
+    additional_log_standard_attrs = []
+    if additional_attrs_str:
+        additional_log_standard_attrs = [attr.strip() for attr in additional_attrs_str.split(",") if attr.strip()]
+
+    # Set log deduplication environment variable if specified
+    dedup_logs = os.environ.get("RAY_DEDUP_LOGS", "1")
+    if dedup_logs is not None:
+        os.environ["RAY_DEDUP_LOGS"] = str(dedup_logs)
+
+    # Set log to driver environment variable if specified
+    log_to_driver = os.environ.get("RAY_LOG_TO_DRIVER", "1")
+    if log_to_driver is not None:
+        os.environ["RAY_LOG_TO_DRIVER"] = str(log_to_driver).lower()
+
+    # Configure log rotation settings
+    rotate_bytes = os.environ.get("RAY_LOGGING_ROTATE_BYTES", "1073741824")  # Default: 1GB per file
+    if rotate_bytes is not None:
+        try:
+            rotate_bytes_int = int(rotate_bytes)
+            os.environ["RAY_LOGGING_ROTATE_BYTES"] = str(rotate_bytes_int)
+        except ValueError:
+            logger.warning(f"Invalid RAY_LOGGING_ROTATE_BYTES '{rotate_bytes}', using default (1GB)")
+            os.environ["RAY_LOGGING_ROTATE_BYTES"] = "1073741824"
+
+    rotate_backup_count = os.environ.get("RAY_LOGGING_ROTATE_BACKUP_COUNT", "19")  # Default: 19 backups (20GB Max)
+    if rotate_backup_count is not None:
+        try:
+            backup_count_int = int(rotate_backup_count)
+            os.environ["RAY_LOGGING_ROTATE_BACKUP_COUNT"] = str(backup_count_int)
+        except ValueError:
+            logger.warning(f"Invalid RAY_LOGGING_ROTATE_BACKUP_COUNT '{rotate_backup_count}', using default (19)")
+            os.environ["RAY_LOGGING_ROTATE_BACKUP_COUNT"] = "19"
+
+    # Configure Ray internal logging verbosity
+    disable_import_warning = os.environ.get("RAY_DISABLE_IMPORT_WARNING", "0")
+    if disable_import_warning is not None:
+        os.environ["RAY_DISABLE_IMPORT_WARNING"] = str(disable_import_warning)
+
+    # Configure usage stats collection
+    usage_stats_enabled = os.environ.get("RAY_USAGE_STATS_ENABLED", "1")
+    if usage_stats_enabled is not None:
+        os.environ["RAY_USAGE_STATS_ENABLED"] = str(usage_stats_enabled)
+
+    # Create LoggingConfig with validated parameters
+    logging_config = LoggingConfig(
+        encoding=encoding,
+        log_level=log_level,
+        additional_log_standard_attrs=additional_log_standard_attrs,
+    )
+
+    logger.info(
+        f"Ray logging configured: preset={preset_level}, level={log_level}, encoding={encoding}, "
+        f"additional_attrs={additional_log_standard_attrs}, "
+        f"dedup_logs={os.environ.get('RAY_DEDUP_LOGS', '1')}, "
+        f"log_to_driver={os.environ.get('RAY_LOG_TO_DRIVER', 'true')}, "
+        f"rotate_bytes={os.environ.get('RAY_LOGGING_ROTATE_BYTES', '1073741824')}, "
+        f"rotate_backup_count={os.environ.get('RAY_LOGGING_ROTATE_BACKUP_COUNT', '19')}"
+    )
+
+    return logging_config
+
+
 def setup_ingestion_pipeline(pipeline: RayPipeline, ingest_config: Dict[str, Any] = None):
     # Initialize the pipeline with the configuration
     if ingest_config:
         # Export the config to environment variables
         export_config_to_env(ingest_config)
 
-    current_level = logging.getLogger().getEffectiveLevel()
+    _ = logging.getLogger().getEffectiveLevel()
+    logging_config = build_logging_config_from_env()
     ray_context = ray.init(
         namespace="nv_ingest_ray",
-        logging_level=current_level,
+        logging_config=logging_config,
         ignore_reinit_error=True,
         dashboard_host="0.0.0.0",
         dashboard_port=8265,

{nv_ingest-2025.8.2.dev20250802 → nv_ingest-2025.8.4.dev20250804}/nv_ingest/framework/orchestration/ray/util/pipeline/stage_builders.py

@@ -242,7 +242,7 @@ def add_table_extractor_stage(pipeline, default_cpu_count, stage_name="table_ext
         stage_actor=TableExtractorStage,
         config=table_extractor_config,
         min_replicas=0,
-        max_replicas=_get_max_replicas(default_cpu_count, percentage_of_cpu=0.20),
+        max_replicas=_get_max_replicas(default_cpu_count, percentage_of_cpu=0.20, replica_limit=4),
     )
 
     return stage_name
@@ -271,7 +271,7 @@ def add_chart_extractor_stage(pipeline, default_cpu_count, stage_name="chart_ext
         stage_actor=ChartExtractorStage,
         config=chart_extractor_config,
         min_replicas=0,
-        max_replicas=_get_max_replicas(default_cpu_count, percentage_of_cpu=0.20),
+        max_replicas=_get_max_replicas(default_cpu_count, percentage_of_cpu=0.20, replica_limit=4),
     )
 
     return stage_name
@@ -417,7 +417,7 @@ def add_otel_tracer_stage(pipeline, default_cpu_count, stage_name="otel_tracer")
         stage_actor=OpenTelemetryTracerStage,
         config=otel_tracer_config,
         min_replicas=0,
-        max_replicas=2,
+        max_replicas=1,
     )
 
     return stage_name
@@ -523,7 +523,7 @@ def add_text_embedding_stage(pipeline, default_cpu_count, stage_name="text_embed
         stage_actor=TextEmbeddingTransformStage,
         config=config,
         min_replicas=0,
-        max_replicas=_get_max_replicas(default_cpu_count, percentage_of_cpu=0.07, replica_limit=6),
+        max_replicas=_get_max_replicas(default_cpu_count, percentage_of_cpu=0.07, replica_limit=4),
     )
 
     return stage_name

{nv_ingest-2025.8.2.dev20250802 → nv_ingest-2025.8.4.dev20250804}/nv_ingest.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nv-ingest
-Version: 2025.8.2.dev20250802
+Version: 2025.8.4.dev20250804
 Summary: Python module for multimodal document ingestion
 Author-email: Jeremy Dyer <jdyer@nvidia.com>
 License:                                  Apache License