Log4j 2 has a BurstFilter that can be used for throttling.

Ralph

> On Mar 7, 2016, at 9:19 AM, Nicholas Duane <nic...@msn.com> wrote:
> 
> I was asking about what's getting logged as I figured we shouldn't have that 
> much in there to be too worried about disk space consumption.  The appender 
> in question is an https appender we wrote.  If it encounters an exception is 
> calls LOGGER.error().  We should have some throttling around that such that 
> we don't generate 1-for-1 errors for each exception we encounter since a 
> problem with the https endpoint could flood the system, but we don't have 
> that throttling yet.  We're redirecting stdout now which is why we're running 
> into some disk consumption issue.  Is there a safe way to implement rolling 
> the file without the generating app knowing about it?  Wouldn't I potentially 
> be deleting the file contents out from under it?
> 
> I have a related question, does log4j2 have some throttling of its own?  If I 
> removed the LOGGER.error() and just let the exception bubble up to log4j2, 
> would it throttle those errors?  I think I was investigating this earlier and 
> noticed some throttling but can't be sure.
> 
> ...<some time going by while I'm thinking>...
> 
> I guess we could write our own application which takes stdin and writes to 
> files and this could do the rolling file work.  I would think something like 
> that might already exist in linux.
> 
> Thanks,
> Nick
> 
>> Subject: Re: controlling the status logger output?
>> From: ralph.go...@dslextreme.com
>> Date: Mon, 7 Mar 2016 08:54:05 -0700
>> To: log4j-user@logging.apache.org
>> 
>> If you set the status level to ERROR the StatusLogger should generate very 
>> little output. That said, if you are concerned you can redirect stdout and 
>> implement the rolling yourself.
>> 
>> Ralph
>> 
>>> On Mar 7, 2016, at 8:39 AM, Nicholas Duane <nic...@msn.com> wrote:
>>> 
>>> We've written some appenders and I think the prescribed approach is to use 
>>> the status logger in log4j2 components, is that correct?  The problem we're 
>>> running into is that we redirect stdout to a file and thus that file can 
>>> grow unbounded.  It seems there's no way to have something like a rolling 
>>> file appender for the status logger, correct?  Any suggestions for limiting 
>>> the size of the log generated by the status logger when stdout has been 
>>> redirected to a file?
>>> 
>>> Thanks,
>>> Nick
>>>                                       
>> 
>> 
>> 
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: log4j-user-unsubscr...@logging.apache.org
>> For additional commands, e-mail: log4j-user-h...@logging.apache.org
>> 
>                                         



---------------------------------------------------------------------
To unsubscribe, e-mail: log4j-user-unsubscr...@logging.apache.org
For additional commands, e-mail: log4j-user-h...@logging.apache.org

Reply via email to