While we're bringing up topics to discuss, I thought I'd throw this one out there. Deploying systems into production (or just into serious test mode) always brings up one issue pretty much consistently. What to do about logging? The same questions/issues always seem to come up:
- How do I know what went wrong and where? - How do I get access to the service logs being created at each individual machine? - Can I trace an invocation across services (that might be on different machines) using log messages, and how? I've transitioned over to SLF4J sometime ago, and found the options available using either Logback or Log4J to completely outshine any capabilities of java.util.logging (j.u.l), and start to make the above questions addressable (without having to write your own customized logging framework). Some of the top advatanges I've found are: - Pick the logging system at deploy tie (Logback,Log4J, j.u.l) - Choice of appenders that are available to do log consolidation & management - Better configuration for logging and loggers - Having mapped diagnostic context available (if the underlying logging system supports it as logback and log4J do) - Performance advantages that logabck and Log4J offer For River to be a key player in the enterprise, it has to be able to use logging framework that allows better management, visibility and control. For that reason I'd like to suggest that we move River away from j.u.l to SLF4J. Regards Dennis
