I've now created my own simple filter which I've added as the first filter to each of my <appender> definitions - it checks for the presence of a predefined marker and always ACCEPTs those messages - giving the behaviour I required.
Regards, Shane On 19 December 2012 00:30, Shane Kelly <[email protected]> wrote: > Hi Daniel, > > Thanks for the info - glad there are others out there with similar > requirements. I guess a separate logger specifically for this purpose is > another way to approach it alright - definitely worth considering. I think > I'm gonna try extending the existing ThresholdFilter (or creating a new > filter based on it) as a first attempt, see how that goes, and failing that > try the separate logger route. Good to have options. > > Regards, > > Shane > > > On 19 December 2012 00:12, Daniel Ferber <[email protected]> wrote: > >> Hi Shane, >> >> In a similar situation, I used another approach. First I considered using >> markers on each initialization log message. Then I used a logger dedicated >> exclusively for initialization. This dedicated logger was configured to log >> everything, regardless of other loggers. >> >> Both solutions, markers or dedicated logger, require you to handle the >> initialization messages differently and to provide proper configuration for >> them. IMHO, the approach of a dedicated logger is much simpler to >> understand and to configure. >> >> Best regards, >> Daniel Felix Ferber >> >> 2012/12/14 Shane Kelly <[email protected]> >> >>> Folks, >>> >>> Just wondering if there is a capability within Logback for writing a log >>> message regardless of whatever log level has been set in configuration. >>> Consider the scenario where I want my web application to output some >>> diagnostic information at startup or shutdown - for example, the Web >>> Application version, build date etc. If I were to set the log level of >>> these messages to be TRACE, DEBUG, or INFO then its possible they may never >>> be displayed since the app may typically be configured to run with a log >>> level of WARN. Similarly, I don't want to set the log level of the messages >>> to WARN, ERROR or FATAL in order to ensure that they do get displayed since >>> they're not really error messages, and if we monitor the log files for >>> WARN, ERROR or FATAL messages then this would trigger a false positive. >>> >>> So, is there some way to force a message to be logged at all times, >>> independently of log level? Or some way to achieve this effect via existing >>> configuration. Arguably I suppose this is bending the rules slightly, in >>> that it could be abused - why offer the ability to filter certain log >>> levels if an application can override/ignore them - but perhaps this is >>> something which could be configurable/switchable? >>> >>> Regards, >>> >>> Shane >>> >>> _______________________________________________ >>> Logback-user mailing list >>> [email protected] >>> http://mailman.qos.ch/mailman/listinfo/logback-user >>> >> >> >> _______________________________________________ >> Logback-user mailing list >> [email protected] >> http://mailman.qos.ch/mailman/listinfo/logback-user >> > >
_______________________________________________ Logback-user mailing list [email protected] http://mailman.qos.ch/mailman/listinfo/logback-user
