Sangeeta, On 8/11/14, 5:41 AM, sangeeta lal wrote: > Actually I have data for other log levels also. Debug =600 statements, > error=400 statements, trace =90 statements etc.
I would usually expect in a typical project that there would be more TRACE logging statements than anything else. On the other hand, DEBUG tends to be the default log level used by most developers that I observe. There are likely many DEBUG statements in Tomcat's code that perhaps should be TRACE statements. 400 ERROR versus 600 DEBUG seems like an awfully large number of ERROR statements, but that may simply be evidence that most errors are properly-logged while there is less DEBUG logging than average. > I am just curious, what could be the possible reason for having such few > "fatal" statements. Can you give your opinion about this? There aren't too many things that ate truly /fatal/ to Tomcat. If we can read config files, mostly everything is okay. One might consider that failing to bind to a port is a fatal error, but Tomcat can start up "successfully" even if no connectors can start properly. This is because connectors can be configured on the fly, etc. and, in embedded contexts, the state of the container can change from within and therefore zero live connectors is no cause for alarm. Most errors don't take-down the container/JVM, so they aren't considered fatal. I wouldn't expect to see very many FATAL log messages in any product, really: the truly fatal things happen at the JVM level and would end up emitting a message to stdout and possibly bringing-down the JVM entirely (e.g. segmentation fault). If you have some /suggestions/ for what conditions might be fatal, we might be able to comment on those specifically. But, we aren't going to re-evaluate every component in Tomcat for logging to satisfy your academic curiosity about logging practices in the Tomcat source. -chris
signature.asc
Description: OpenPGP digital signature
