geronimo-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jason Dillon <>
Subject Re: svn commit: r525594 - /geronimo/server/trunk/assemblies/geronimo-boilerplate-minimal/src/main/resources/var/log/
Date Thu, 05 Apr 2007 02:38:20 GMT
On Apr 4, 2007, at 7:31 PM, Lin Sun wrote:
> I proposed to add them in just because it took me quite a while to  
> figure out how to turn on debug logs for Axis2.  This is needed if  
> anyone ever needs to debug an Axis2 prob or report an Axis2 JIRA.
> I don't want others to go through the effort thus I feel it is  
> appropriate to have them in so that people can change to DEBUG or  
> TRACE whenever needed.
> It is perfectly fine if you want to have INFO or ERROR instead of  
> FATAL.  To answer your question, I don't really see much/any output  
> of these loggers w/o the limits.

Wiki pages are quite useful for explain these types of things to  
people.  If there isn't already there should be a comprehensive guide  
on how to deal with logging in Geronimo.  I don't think it is  
appropriate to use the actually logging configuration file as  
documentation... though these aren't commented out with docs, they  
are actually setting the levels which I think is even worse.

I would recommend creating that wiki page (or updating it if one  
exists, I've not looked recently at what is there).

And if you want, add _commented_ examples of how to enable DEBUG/ 
TRACE, but don't limit things that aren't insanely noisy by default.   
That will only leave folks scratching their heads when they expect to  
see the output.

And I would recommend that we should never be limiting categories to  
FATAL or ERROR... those log messages generally indicate problems  
which should not be swallowed by default.


> Lin
> Jason Dillon wrote:
>> I don't think we should be trying to taylor the logging output of  
>> Geronimo to match what other component communities have for their  
>> projects.
>> I'm okay with with levels for axis, though I think FATAL is not  
>> the correct level to limit them at by default.  In most cases I  
>> would expect to see INFO+ captured in log files, but when limiting  
>> logger levels like this they will never make it to the file appender.
>> But the others, like httpclient for example.  Users may be using  
>> Geronimo w/o any WS muck, using httpclient and expecting to see  
>> log messages.  Limiting these logger here is a very bad idea, as  
>> it will leave those users wonder where the logs went and causing  
>> them to ping the lists asking what is going on.
>>  * * *
>> In general I'm -1 on limiting loggers to FATAL, unless for some  
>> reason the component spits out a ton of ERROR messages, and  
>> similarly I'm -1 on limiting loggers to ERROR unless they spit out  
>> a ton of WARN messages.  And in both cases if those libraries are  
>> spitting out so much junk, then we are either integrating them  
>> improperly or their codebase is incorrectly using logging... in  
>> both cases something should be fixed, we shouldn't be silently  
>> ignoring them.
>>  * * *
>> What is the output of these loggers w/o the limits?
>> --jason
>> On Apr 4, 2007, at 1:39 PM, Donald Woods wrote:
>>> The used by Axis2 includes those 4 values set to  
>>> FATAL, so I'm trying to match our log output to what the Axis2  
>>> community is used to seeing and ships today.
>>> Also, adding these values into our file allows users to easily  
>>> see how to turn on debug info for a component (Axis and Axis2 in  
>>> this case) without having to dig through the component source or  
>>> pinging our user mailing list for the info.....
>>> -Donald
>>> Jason Dillon wrote:
>>>> Why?
>>>> I don't think its a good idea to keep growing the list of logger  
>>>> levels in our log4j configuration file like this.  For one or  
>>>> two its okay, but probably not for so many.  I mean, do these  
>>>> libraries really spit out so much information that we have to  
>>>> limit them all to FATAL?
>>>> The default output level is currently set to WARN unless the -v  
>>>> or -vv flag is passed to the server, which will set to DEBUG and  
>>>> TRACE respectively.  With logger levels set explicitly , then  
>>>> adding -v or -vv will have zero affect.  And the way we  
>>>> currently configure these levels affect both the console and log  
>>>> files.
>>>> I think that changing these levels to FATAL is harmful and  
>>>> should be reverted... unless there is a really good reason for  
>>>> it... which is what I'm asking right now.  What is the reason we  
>>>> need to have these explicit logger levels configured here?
>>>> --jason
>>>> On Apr 4, 2007, at 1:05 PM, wrote:
>>>>> Author: dwoods
>>>>> Date: Wed Apr  4 13:05:30 2007
>>>>> New Revision: 525594
>>>>> URL:
>>>>> Log:
>>>>> GERONIMO-3064 Add axis2 log4j configure properties so that  
>>>>> people can turn on axis2 logs in geronimo.  Thanks Lin.  I also  
>>>>> added the Axis v1 log categories.
>>>>> Modified:
>>>>>     geronimo/server/trunk/assemblies/geronimo-boilerplate- 
>>>>> minimal/src/main/resources/var/log/
>>>>> Modified: geronimo/server/trunk/assemblies/geronimo-boilerplate- 
>>>>> minimal/src/main/resources/var/log/
>>>>> URL: 
>>>>> assemblies/geronimo-boilerplate-minimal/src/main/resources/var/ 
>>>>> log/ 
>>>>> view=diff&rev=525594&r1=525593&r2=525594
>>>>> ================================================================== 
>>>>> ============
>>>>> --- geronimo/server/trunk/assemblies/geronimo-boilerplate- 
>>>>> minimal/src/main/resources/var/log/  
>>>>> (original)
>>>>> +++ geronimo/server/trunk/assemblies/geronimo-boilerplate- 
>>>>> minimal/src/main/resources/var/log/ Wed  
>>>>> Apr  4 13:05:30 2007
>>>>> @@ -115,10 +115,22 @@
>>>>>  # Prints various stuff during startup
>>>>> -
>>>>>  # Prints various stuff when the portal is used
>>>>> +
>>>>>  # Prints stuff for AJAX calls
>>>>> +
>>>>> +# Axis log output
>>>>> +
>>>>> +# Axis2 log output
>>>>> +log4j.logger.httpclient.wire.header=FATAL
>>>>> +

View raw message