Table of Contents
Trac supports logging of system messages using Python's logging module.
Supported Logging Methods
The log method is set using the [logging] log_type option, which accepts the following values:
- Suppress all log messages.
- Log messages to a file, specified with the [logging] log_file option. Relative paths for
log_fileare resolved relative to the environment
- Output log entries to console (tracd only).
- (UNIX) Send log messages to the local syslogd via named pipe
/dev/log. By default, syslog will write them to the
- (Windows) Use the system's NT Event Log.
The verbosity level of logged messages can be set using the [logging] log_level option. The log level defines the minimum level of urgency required for a message to be logged, and those levels are:
- Log only the most critical (typically fatal) errors.
- Log failures, bugs and errors.
- Log warnings, non-interrupting events.
- Diagnostic information, log information about all processing.
- Trace messages, profiling, etc.
Additionally, you can enable logging of SQL statements at debug level. This is turned off by default, as it's very verbose. Set [trac] debug_sql = yes to activate.
The output format for log entries can be specified in the [logging] log_format option. The format is a string which can contain any of the Python logging Formatter variables. Additionally, the following Trac-specific variables can be used:
- The environment directory name.
- The absolute path of the environment directory.
- The originating project's name.
Note that variables are identified using a dollar sign (
$(...)s) instead of percent sign (
The default format is:
log_format = Trac[$(module)s] $(levelname)s: $(message)s
In a multi-project environment where all logs are sent to the same place (e.g.
syslog), it makes sense to add the project name. In this example we use
basename since that can generally be used to identify a project:
log_format = Trac[$(basename)s:$(module)s] $(levelname)s: $(message)s