np_logging

For use on internal Allen Institute network

Quick start

Follow the conventions for using the logging module in the standard library:

• in a script or main module, np_logging adds console & file handlers and exit messages to the root logger:

import np_logging

logger = np_logging.getLogger()

• or, in a package or project with multiple modules, all logged messages are propagated to the root logger:

import np_logging

logger = np_logging.getLogger(__name__)     # logger.level = logging.NOTSET = 0

• then log messages as usual:

logger.info('test message')
logger.warning('test message')

No further setup is required, and importing logging from the standard library isn't necessary.

---

To send a message to the Mindscope log-server, use np_logging.web() and supply a project name, which will appear in the channel field on the server:

project_name = 'spike_sorting'

np_logging.web(project_name).info('test message')

• the web log can be viewed at http://eng-mindscope:8080

---

For customization, use np_logging.setup() to supply a logging config dict that specifies loggers and their handlers & formatters, and np_logging will add extra functionality such as exit messages/emails.

• logging configs should be specified according to the python logging library dict schema

• logging configs on the eng-mindscope ZooKeeper server can also be used directly to setup logging by supplying their path to np_logging.setup():

np_logging.setup(
    '/projects/np_workflows/defaults/logging'
)

See np_config for further info on using ZooKeeper for configs.

Other input arguments to np_logging.setup():

• project_name (default: current working directory name)

  • sets the channel displayed on the log server

• email_address (default: None)

  • if one or more addresses are supplied, an email is sent at program exit reporting the elapsed time and cause of termination. If an exception was raised, the traceback is included.

• log_at_exit (default: True)

  • If True, a message is logged when the program terminates, reporting total elapsed time.

• email_at_exit (default: False, or True if email_address is not None)

  • If True, an email is sent when the program terminates.

  • If logging.ERROR, the email is only sent if the program terminates via an exception.