How will you investigate a failure that users encounter but you cannot reproduce? Record application events as they happen. Logs preserve clues about what the program did, making reported problems easier to diagnose. Add logging during development rather than waiting for a production failure. section 12.0 section 12.1
Keep an error record
import logging logger = logging.getLogger(__name__) file_handler = logging.FileHandler("taskier.log") logger.addHandler(file_handler) logger.error("Failed to save the task to the db") The file now contains: Failed to save the task to the db. Unlike console output from print, the record remains available after the console closes. Retrieving the logger by module name also avoids creating separate loggers for that module. section 12.1.2 section 12.2.3 section 12.6
Configure the logger to keep useful records without burying important ones:
- Add a StreamHandler alongside the FileHandler when you also want real-time console feedback during development; both handlers can receive the same event. section 12.1.3
- Choose among DEBUG, INFO, WARNING, ERROR, and CRITICAL by severity. A logger set to WARNING captures WARNING and more serious records, but not INFO or DEBUG. section 12.2.1
- Set levels on individual handlers when destinations need different records. Keep the logger’s threshold low enough to admit anything those handlers should receive. section 12.2.2 section 12.2.4
- Attach a Formatter to make each record easier to find and interpret. A timestamp, severity, module name, and message provide more context than a bare message. section 12.2.3
Listing 12.6 Setting levels to individual file handlers
logger.setLevel(logging.DEBUG) #1
handler_warning = logging.FileHandler("taskier_warning.log")
handler_warning.setLevel(logging.WARNING) #2
logger.addHandler(handler_warning)
handler_critical = logging.FileHandler("taskier_critical.log")
handler_critical.setLevel(logging.CRITICAL) #3
logger.addHandler(handler_critical)
logging_messages_all_levels()
warning_log_records = check_log_content("taskier_warning.log")
print(warning_log_records)
# output the following lines:
--Critical message
--Error message
--Warning message
critical_log_records = check_log_content("taskier_critical.log")
print(critical_log_records)
# output the following line:
--Critical message
Compare the two output files: the WARNING handler keeps warnings, errors, and critical messages, while the CRITICAL handler keeps only critical messages. section 12.2.2
For ongoing monitoring, save important events to a file; use console output as a convenient development aid. Format records and choose levels deliberately so a large log remains useful when a problem appears. section 12.1.4 section 12.2
