Hello Rainer, Thanks detailed explain. and now I use jconsole to measuring. But when jconsole try to connect remote target host, JVM down. Probably it is not tomcat issue,I think. I may close this issue. Regard,
--catalina.out-- ... # # An unexpected error has been detected by Java Runtime Environment: # # Internal Error (os_linux_zero.cpp:157), pid=1696, tid=1735849104 # Error: caught unhandled signal 11 # # Java VM: OpenJDK Core VM (1.6.0_0-b11 interpreted mode linux-arm) # An error report file with more information is saved as: # /root/hs_err_pid1696.log # # If you would like to submit a bug report, please visit: # http://icedtea.classpath.org/bugzilla # The crash happened outside the Java Virtual Machine in native code. # See problematic frame for where to report the bug. --- On 31 August 2010 20:57, Rainer Jung <rainer.j...@kippdata.de> wrote: > On 31.08.2010 13:14, takanobu watanabe wrote: >> >> Thank you very much Rainer. >> >>> Tomcat itself doesn't need much memory and you should be able to run it >>> with less than 64MB Heap (plus Perm) >> >> I recognized 64MB>= Am + Bm from your advice. >> Is my recognized correct ? >> ---example catalina.sh--- >> -XX:NewSize=Am -XX:MaxNewSize=Am >> -XX:PermSize=Bm -XX:MaxPermSize=Bm" > > I don't actually understand the question, but here's the terminology (I hope > the picture isn't disturbed by line breaks): > > <---------------------- Heap -------------------><- Perm -> > <----------------- New ----------------><- Old -> > <- Eden -><- Survivor1 -><- Survivor2 -> > > Sometimes New is also defined as Eden plus one of the two survivors, > especially in the verbose GC output. > > The two survivors are also called "From Space" and "To Space", the roles > change during each GC run in New. > > Xms, Xmx: minimum and maximum heap size > XX:NewSize, XX:MaxNewSize: minimum and maximum size of New > XX:PermSize and XX:MaxPermSize: minimum and maximum size of Perm > > Sizes of Survivor spaces (each one) are determined by > > Size of Survivor = Size of New / (SurvivorRatio + 2) > > So if SurvivorRatio is some number N, then the Ratios are > > Eden:Survivor1:Survivor2 = N:1:1 > > and New = (N+2) * One Survivor > > Regards, > > Rainer > >> On 31 August 2010 19:00, Rainer Jung<rainer.j...@kippdata.de> wrote: >>> >>> On 31.08.2010 10:34, takanobu watanabe wrote: >>>> >>>> Hello, >>>> >>>> What is default value of CATALINA_OPTS at apache-tomcat-6.0.29 ? >>>> It means No configure catalina.sh as same as value of CATALINA_OPTS >>>> just expand tomcat. >>>> >>>> Because I try to run tomcat lean physical memory condition(128Mb only >>>> with armv5). >>>> When nothing configure to catalina.sh(default),Tomcat works well. >>>> But when I configure below,looks like always FullGC running and could >>>> not access WEB service. >>>> --- >>>> CATALINA_OPTS=" -verbose:gc -XX:+PrintGCTimeStamps -Xms64m -Xmx64m >>>> -XX:NewSize=32m -XX:MaxNewSize=32m >>>> -XX:SurvivorRatio=2 -XX:PermSize=16m -XX:MaxPermSize=16m" >>>> --- >>>> If anyone has idea or suggest of run tomcat on lean physical memory >>>> condition, I would like to know. >>> >>> If you don't add explicit memory options, the default memory sizes of >>> your >>> JVM are used. Those might depend on the exact jvm version and also on the >>> type of system you run it on (total memory, CPUs). >>> >>> The easiest way to find out the sizes for your jvm are running it without >>> explicit settings and then measuring it. Measuring can be done with tools >>> like jstat, jconsole or jvisualvm, or you can add a few verbose gc flags >>> to >>> print out the sizes of the different memory regions during each run of >>> the >>> GC, e.g. >>> >>> -verbose:gc -XX:+PrintGCDetails -XX:+PrintHeapAtGC -XX:+PrintGCTimeStamps >>> >>> Note that the default settings will allow some of the regions to grow and >>> shrink, so you need to make observations for a longer time to find out >>> the >>> values which are effective most of the time. >>> >>> I expect your above settings might not leave enough room in the Old >>> Generation (aka Tenured), because your New Generation is somwhat large >>> relative to the Old Gen. Tomcat itself doesn't need much memory and you >>> should be able to run it with less than 64MB Heap (plus Perm), but what >>> you >>> need in the end depends on the application inside (and the concurrency >>> induced by the load). I never experimented to find a minimum value >>> though. >>> >>> Regards, >>> >>> Rainer > > --------------------------------------------------------------------- > To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org > For additional commands, e-mail: users-h...@tomcat.apache.org > > -- tknv/ --------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org
