Add formatting to log and update get_logger function. (#1204)

Co-authored-by: jlewitt1 <josh@run.house>

examples/airflow-torch-training/callables.py

@@ -4,7 +4,7 @@
 from torch_example_for_airflow import DownloadData, SimpleTrainer
 
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 def bring_up_cluster_callable(**kwargs):

runhouse/logger.py

@@ -1,15 +1,59 @@
 import logging
 import os
+import sys
 
 
-def get_logger(name: str = __name__):
-    logger = logging.getLogger(name)
+def get_logger(reinitialize: bool = False) -> logging.Logger:
+    """
+    Creates and returns a logger with the specified name.
 
-    level = os.getenv("RH_LOG_LEVEL")
-    if level:
-        try:
-            logger.setLevel(getattr(logging, level.upper()))
-        except AttributeError as e:
-            raise e
+    Ensures a universal logger configuration across the codebase with the format:
+    "levelname - asctime - filename:lineno - message"
+
+    Args:
+        name (str): Name of the logger. Defaults to None, which gets the root logger.
+
+    Returns:
+        logging.Logger: Configured logger instance.
+    """
+    # Create or retrieve the logger
+    logger = logging.getLogger("_rh_universal_logger")
+
+    level = os.getenv("RH_LOG_LEVEL") or "INFO"
+    try:
+        level = getattr(logging, level.upper())
+    except AttributeError as e:
+        raise e
+
+    if reinitialize:
+        for handler in logger.handlers:
+            logger.removeHandler(handler)
+
+    # Check if the logger has already been configured to prevent duplicate handlers
+    if not logger.handlers:
+
+        # Set the level for the new logger
+        logger.setLevel(level)
+
+        # Create a console handler that outputs to stdout
+        console_handler = logging.StreamHandler(stream=sys.stdout)
+
+        # The handler should capture all log levels, so we set the level to debug
+        console_handler.setLevel(logging.DEBUG)
+
+        # Define the log format
+        formatter = logging.Formatter(
+            "%(levelname)s | %(asctime)s | %(filename)s:%(lineno)d | %(message)s",
+            datefmt="%Y-%m-%d %H:%M:%S",
+        )
+
+        # Apply the formatter to the handler
+        console_handler.setFormatter(formatter)
+
+        # Add the handler to the logger
+        logger.addHandler(console_handler)
+
+        # Prevent log messages from propagating to the root logger
+        logger.propagate = False
 
     return logger

runhouse/main.py

@@ -46,7 +46,7 @@
 # For printing with typer
 console = Console()
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 @app.command()

runhouse/resources/envs/conda_env.py

@@ -14,7 +14,7 @@
 
 from .env import Env
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 class CondaEnv(Env):

runhouse/resources/envs/env.py

@@ -13,7 +13,7 @@
 from runhouse.resources.resource import Resource
 from runhouse.utils import run_with_logs
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 class Env(Resource):

runhouse/resources/folders/folder.py

@@ -14,7 +14,7 @@
 from runhouse.rns.utils.api import generate_uuid, relative_file_path
 from runhouse.utils import locate_working_dir
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 class Folder(Resource):

runhouse/resources/folders/folder_factory.py

@@ -6,7 +6,7 @@
 from runhouse.resources.folders.folder import Folder
 from runhouse.resources.hardware.utils import _get_cluster_from
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 def folder(

runhouse/resources/folders/gcs_folder.py

@@ -8,7 +8,7 @@
 
 from .folder import Folder
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 class GCSFolder(Folder):

runhouse/resources/folders/s3_folder.py

@@ -13,7 +13,7 @@
 POLL_INTERVAL = 1
 TIMEOUT_SECONDS = 3600
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 class S3Folder(Folder):

runhouse/resources/functions/aws_lambda.py

@@ -25,7 +25,7 @@
 CRED_PATH = f"{Path.home()}/.aws/credentials"
 LOG_GROUP_PREFIX = "/aws/lambda/"
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 class LambdaFunction(Function):

runhouse/resources/functions/function.py

@@ -11,7 +11,7 @@
 
 from runhouse.resources.resource import Resource
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 class Function(Module):

runhouse/resources/functions/function_factory.py

@@ -8,7 +8,7 @@
 from runhouse.resources.functions.function import Function
 from runhouse.resources.packages import git_package
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 def function(

runhouse/resources/hardware/cluster.py

@@ -55,7 +55,7 @@
 
 from runhouse.servers.http import HTTPClient
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 class Cluster(Resource):

runhouse/resources/hardware/cluster_factory.py

@@ -15,7 +15,7 @@
 from .on_demand_cluster import OnDemandCluster
 from .sagemaker.sagemaker_cluster import SageMakerCluster
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 # Cluster factory method
 def cluster(

runhouse/resources/hardware/on_demand_cluster.py

@@ -31,7 +31,7 @@
 
 from .cluster import Cluster
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 class OnDemandCluster(Cluster):

runhouse/resources/hardware/ray_utils.py

@@ -6,7 +6,7 @@
 
 from runhouse.logger import get_logger
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 def check_for_existing_ray_instance():

runhouse/resources/hardware/sagemaker/sagemaker_cluster.py

@@ -44,7 +44,7 @@
 from runhouse.utils import generate_default_name
 
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 ####################################################################################################
 # Caching mechanisms for SSHTunnelForwarder
 ####################################################################################################

runhouse/resources/hardware/sky/command_runner.py

@@ -22,7 +22,7 @@
 ##### RH modification #####
 from runhouse.logger import get_logger
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 ##### RH modification #####
 
 

runhouse/resources/hardware/sky/subprocess_utils.py

@@ -5,7 +5,7 @@
 
 from runhouse.logger import get_logger
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 class CommandError(Exception):
     """Raised when a command fails.

runhouse/resources/hardware/sky_ssh_runner.py

@@ -25,7 +25,7 @@
 )
 
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 # Get rid of the constant "Found credentials in shared credentials file: ~/.aws/credentials" message
 try:

runhouse/resources/module.py

@@ -52,7 +52,7 @@
     "_dumb_signature_cache",
 ]
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 class Module(Resource):

runhouse/resources/packages/package.py

@@ -24,7 +24,7 @@
 
 INSTALL_METHODS = {"local", "reqs", "pip", "conda", "rh"}
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 class CodeSyncError(Exception):

runhouse/resources/provenance.py

@@ -20,7 +20,7 @@
 from runhouse.rns.utils.api import log_timestamp, resolve_absolute_path
 from runhouse.utils import StreamTee
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 class RunStatus(str, Enum):

runhouse/resources/resource.py

@@ -19,7 +19,7 @@
     ResourceVisibility,
 )
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 class Resource:

runhouse/resources/secrets/provider_secrets/ssh_secret.py

@@ -10,7 +10,7 @@
 from runhouse.resources.hardware.cluster import Cluster
 from runhouse.resources.secrets.provider_secrets.provider_secret import ProviderSecret
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 class SSHSecret(ProviderSecret):

runhouse/resources/secrets/secret.py

@@ -14,7 +14,7 @@
 from runhouse.rns.utils.api import load_resp_content, read_resp_data
 from runhouse.utils import generate_default_name
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 class Secret(Resource):

runhouse/resources/secrets/utils.py

@@ -11,7 +11,7 @@
 USER_ENDPOINT = "user/secret"
 
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 def load_config(name: str, endpoint: str = USER_ENDPOINT):

runhouse/rns/defaults.py

@@ -16,7 +16,7 @@
 
 req_ctx = contextvars.ContextVar("rh_ctx", default={})
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 class Defaults:

runhouse/rns/login.py

@@ -7,7 +7,7 @@
 from runhouse.globals import configs, rns_client
 from runhouse.logger import get_logger
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 def is_interactive():

runhouse/rns/rns_client.py

@@ -23,7 +23,7 @@
 
 from runhouse.utils import locate_working_dir
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 # This is a copy of the Pydantic model that we use to validate in Den

runhouse/rns/top_level_rns_fns.py

@@ -8,7 +8,7 @@
 from runhouse.logger import get_logger
 from runhouse.servers.obj_store import ClusterServletSetupOption
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 logging.getLogger("numexpr").setLevel(logging.WARNING)
 

runhouse/servers/autostop_helper.py

@@ -3,7 +3,7 @@
 
 from runhouse.logger import get_logger
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 class AutostopHelper:

runhouse/servers/caddy/config.py

@@ -5,7 +5,7 @@
 from runhouse.constants import DEFAULT_SERVER_PORT
 from runhouse.logger import get_logger
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 SYSTEMCTL_ERROR = "systemctl: command not found"
 
 

runhouse/servers/cluster_servlet.py

@@ -27,7 +27,7 @@
 from runhouse.servers.http.auth import AuthCache
 from runhouse.utils import ColoredFormatter, sync_function
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 class ClusterServletError(Exception):

runhouse/servers/env_servlet.py

@@ -16,7 +16,7 @@
 from runhouse.servers.obj_store import ClusterServletSetupOption
 from runhouse.utils import arun_in_thread, get_node_ip
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 def error_handling_decorator(func):

runhouse/servers/http/auth.py

@@ -5,7 +5,7 @@
 from runhouse.rns.utils.api import load_resp_content, ResourceAccess
 from runhouse.servers.http.http_utils import username_from_token
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 class AuthCache:

runhouse/servers/http/certs.py

@@ -12,7 +12,7 @@
 
 from runhouse.rns.utils.api import resolve_absolute_path
 
-logger = get_logger(name=__name__)
+logger = get_logger()
 
 
 class TLSCertConfig: