Another possible way to investigate is to wrap your RollingFile appender in an Async appender. Since there is only one thread that outputs, if it is a thread local related issue, it should disappear (and it may be use as a workaround until the next release)
2016-07-02 16:10 GMT+02:00 Remko Popma <remko.po...@gmail.com>: > This issue may already have been fixed (see > https://issues.apache.org/jira/browse/LOG4J2-1434 ). > The fix will be in the upcoming 2.6.2 release. > > Can you verify if the problem still occurs with the current Log4j code in > git master? > Instructions on how to check out the latest source are here > <http://logging.apache.org/log4j/2.x/source-repository.html>. After that > you can build the binaries with the command > mvn clean install > > (Or mvn -DskipTests=true clean install if you are in a hurry.) > > > On Sat, Jul 2, 2016 at 5:24 PM, Enric Jaen <enricj...@yahoo.es.invalid> > wrote: > > > Sorry typo again, was 3GB (not 3MB) the total consumed memory by several > > char[]. For example I reveived a 60MB utf8 json and I see it resides > inside > > a 250MB char[] > > I have attached a dump screenshot > > > > Not sure if is relevant but I use RollingFile : > > > > <RollingFile name="PROJECT" > > fileName="${sys:catalina.home}/logs/@webapp.name@.log" > > filePattern="${sys:catalina.home}/logs/@webapp.name@ > .%d{yyyy-MM-dd}.log"> > > > > <PatternLayout pattern="%d{ABSOLUTE} %-5p %mdc{ApiKeyOrEmail}-%mdc{tid} > > [%c{1}] %m%n"/> > > <filters> > > <ThresholdFilter level="@log.level.project@" onMatch="NEUTRAL" > > onMismatch="DENY"/> > > <MarkerFilter marker="API_RESPONSE" onMatch="DENY" > > onMismatch="ACCEPT"/> > > </filters> > > <TimeBasedTriggeringPolicy/> > > </RollingFile> > > > > I will try to reproduce if needed > > > > > > > > El Sábado 2 de julio de 2016 5:28, Matt Sicker <boa...@gmail.com> > > escribió: > > > > > > That's the bug I was trying to find. It sounds related. > > > > On 1 July 2016 at 23:09, Ralph Goers <ralph.go...@dslextreme.com> wrote: > > > > > Could this be the same issue as LOG4J2-1434? > > > > > > Ralph > > > > > > > On Jul 1, 2016, at 6:10 PM, Remko Popma <remko.po...@gmail.com> > wrote: > > > > > > > > Yes, Java 7 or higher, so Java 8 is fine. > > > > > > > > I'm still trying to understand the problem. I understand that the > > > application receives and logs large JSON Strings. > > > > After that I don't understand. > > > > Are these strings 3MB long, and they cause Log4j to build up 250MB > > > buffers? > > > > > > > > Do you have a small program that demonstrates the problem? > > > > > > > > Remko > > > > Sent from my iPhone > > > > > > > >> On 2016/07/02, at 1:45, Enric Jaen <enricj...@yahoo.es.INVALID> > > wrote: > > > >> > > > >> We use Java8 , is that ok?The problem is that we receive large json > > > responses and log4j is creating char[]'s of 250MB up a total of 3MB and > > we > > > are suffering of continuos full gc. By releasing that mem we would go > > > fine.Enric > > > >> > > > >> > > > >> > > > >> El Viernes 1 de julio de 2016 16:50, Remko Popma < > > > remko.po...@gmail.com> escribió: > > > >> > > > >> > > > >> Log4j 2.6.1 needs Java 7. Please provide more details about the > > memory > > > problem. > > > >> > > > >> Sent from my iPhone > > > >> > > > >>> On 2016/07/02, at 0:38, Enric Jaen <enricj...@yahoo.es.INVALID> > > wrote: > > > >>> > > > >>> Hello, > > > >>> In our tomcat application we have a memory problem with logs kept > in > > > heap.I tried using log4j 1.6 and log4j2.enable.threadlocals = false but > > > without effect, what can be the reason? > > > >>> RegardsEnric > > > >> > > > >> > --------------------------------------------------------------------- > > > >> 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 > > > > > > > > > > > > > > > > > > > > --------------------------------------------------------------------- > > > To unsubscribe, e-mail: log4j-user-unsubscr...@logging.apache.org > > > For additional commands, e-mail: log4j-user-h...@logging.apache.org > > > > > > > > > > > > > > -- > > Matt Sicker <boa...@gmail.com > > > > > > > > > > > > > --------------------------------------------------------------------- > > To unsubscribe, e-mail: log4j-user-unsubscr...@logging.apache.org > > For additional commands, e-mail: log4j-user-h...@logging.apache.org > > >
