Configuring Logging

Jupyter Server (and Jupyter Server extension applications such as Jupyter Lab) are Traitlets applications.

By default Traitlets applications log to stderr. You can configure them to log to other locations e.g. log files.

Logging is configured via the logging_config “trait” which accepts a logging.config.dictConfig() object. For more information look for Application.logging_config in Config file and command line options.

Examples

Jupyter Server

A minimal example which logs Jupyter Server output to a file:

c.ServerApp.logging_config = {
    "version": 1,
    "handlers": {
        "logfile": {
            "class": "logging.FileHandler",
            "level": "DEBUG",
            "filename": "jupyter_server.log",
        },
    },
    "loggers": {
        "ServerApp": {
            "level": "DEBUG",
            "handlers": ["console", "logfile"],
        },
    },
}

Note

To keep the default behaviour of logging to stderr ensure the console handler (provided by Traitlets) is included in the list of handlers.

Warning

Be aware that the ServerApp log may contain security tokens. If redirecting to log files ensure they have appropriate permissions.

Configuring Log Scrubbing

By default, Jupyter Server scrubs sensitive URL parameters from log output to prevent security tokens and other sensitive information from being leaked in log files. Additional parameters to be scrubbed can be configured using the extra_log_scrub_param_keys trait.

Default scrubbed parameter keys include: token, auth, key, code, state, and xsrf.

Example configuration to add additional parameters to scrub:

# jupyter_server_config.py

# Add additional parameter keys to scrub (these will be added to the defaults)
c.ServerApp.extra_log_scrub_param_keys = [
    "password", "secret", "api_key", "jwt-token"
]

Jupyter Server Extension Applications (e.g. Jupyter Lab)

An example which logs both Jupyter Server and Jupyter Lab output to a file:

Note

Because Jupyter Server and its extension applications are separate Traitlets applications their logging must be configured separately.

c.ServerApp.logging_config = {
    "version": 1,
    "handlers": {
        "logfile": {
            "class": "logging.FileHandler",
            "level": "DEBUG",
            "filename": "jupyter_server.log",
            "formatter": "my_format",
        },
    },
    "formatters": {
        "my_format": {
            "format": "%(asctime)s %(levelname)-8s %(name)-15s %(message)s",
            "datefmt": "%Y-%m-%d %H:%M:%S",
        },
    },
    "loggers": {
        "ServerApp": {
            "level": "DEBUG",
            "handlers": ["console", "logfile"],
        },
    },
}

c.LabApp.logging_config = {
    "version": 1,
    "handlers": {
        "logfile": {
            "class": "logging.FileHandler",
            "level": "DEBUG",
            "filename": "jupyter_server.log",
            "formatter": "my_format",
        },
    },
    "formatters": {
        "my_format": {
            "format": "%(asctime)s %(levelname)-8s %(name)-15s %(message)s",
            "datefmt": "%Y-%m-%d %H:%M:%S",
        },
    },
    "loggers": {
        "LabApp": {
            "level": "DEBUG",
            "handlers": ["console", "logfile"],
        },
    },
}

Note

The configured application name should match the logger name e.g. c.LabApp.logging_config defines a logger called LabApp.

Tip

This diff modifies the example to log Jupyter Server and Jupyter Lab output to different files:

--- before
+++ after
 c.LabApp.logging_config = {
     'version': 1,
     'handlers': {
         'logfile': {
             'class': 'logging.FileHandler',
             'level': 'DEBUG',
-            'filename': 'jupyter_server.log',
+            'filename': 'jupyter_lab.log',
             'formatter': 'my_format',
         },
     },