commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Mark Thomas <ma...@apache.org>
Subject Re: [DISCUSS][POOL] Logging options for Pool2
Date Sat, 18 Feb 2012 11:11:26 GMT
On 17/02/2012 21:13, Gary Gregory wrote:
> So... are we ready to move some bits?
> ;)

Getting there. I had some Tomcat stuff to do. I'll try and look at this
this week.

Mark

> Gary
> 
> On Sun, Feb 12, 2012 at 8:03 PM, Konstantin Kolinko
> <knst.kolinko@gmail.com>wrote:
> 
>> 2012/2/13 Mark Thomas <markt@apache.org>:
>>>
>>> General
>>> - Logging in pool, if any, should be minimal
>>>
>>
>> Two general questions:
>> When there are several pools,
>> - is it possible to discern log messages from different pools?
>> - is it possible to control logging level for a single pool, or all
>> pools have the same logging configuration? E.g. to enable debug
>> logging for a single pool only (if debug logging is ever implemented
>> there).
>>
>>>
>>> What do folks think to the following solution:
>>> - make logging of factory exceptions a factory responsibility - clearly
>>> documented in the JavaDoc for POOL2
>>> - add some JMX stats to POOL2 for number of destroy / passivate /
>>> activate exceptions
>>> - make the last n (10?) exceptions of each type accessible via JMX for
>> POOL2
>>> ?
>>>
>>
>> Overall this proposal looks good, but there is known caveat with
>> keeping Exception instances in memory:  Their stacktrace data is known
>> to pin classloaders in memory, preventing them from being garbage
>> collected and thus causing PermGen memory leaks in web applications.
>>
>> This problem was reproduced in this Tomcat issue:
>> https://issues.apache.org/bugzilla/show_bug.cgi?id=50460
>> Discussion:
>> http://marc.info/?l=tomcat-dev&m=129211856426188&w=2
>>
>> A workaround might be to print stacktraces as strings and keep the
>> strings in memory, but it may be time-consuming. The good point though
>> is that the time to print those exceptions to strings is comparable
>> with the one spent when printing them to a log file.
>>
>> Best regards,
>> Konstantin Kolinko
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: dev-unsubscribe@commons.apache.org
>> For additional commands, e-mail: dev-help@commons.apache.org
>>
>>
> 
> 


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@commons.apache.org
For additional commands, e-mail: dev-help@commons.apache.org


Mime
View raw message