Hello, Please, provide additional information. Can you provide thread dump from server? What processes doing on cluster?
Need to understanding, where garbage is generated. On Tue, Aug 16, 2016 at 12:39 PM, yucigou <[email protected]> wrote: > I'm a bit puzzled about the basic JVM configuration (which is said has > proven > to provide fairly smooth throughput without large spikes) on the Ignite JVM > and System Tuning page > (https://apacheignite.readme.io/docs/jvm-and-system-tuning): > > -XX:NewSize=128m > -XX:MaxNewSize=128m > -XX:MaxTenuringThreshold=0 > -XX:SurvivorRatio=1024 > > Below is the CPU and memory profile captured: > <http://apache-ignite-users.70518.x6.nabble.com/file/ > n7091/Ignite_basic_JVM_configuration.png> > > (By the way, there is no heavy job running for my application, except for > the Ignite REST API enabled with Ignite web console invoking it through it > agent.) > > You can see that the minor GCs are so frequent, and also the full GCs, and > the CPU usage is so high as result. The size of young gen space is very > small. Also live objects are not aged at all, and they are promoted to old > gen immediately. > > I just wonder what is the reasoning behind these settings: > (1) Small young gen space size > (2) Tiny survivor space with max tenuring threshold set to zero (which > means > no aging) > > Thank you, > Yuci > > > > > > > -- > View this message in context: http://apache-ignite-users. > 70518.x6.nabble.com/Puzzled-about-JVM-configuration-tp7091.html > Sent from the Apache Ignite Users mailing list archive at Nabble.com. > -- Vladislav Pyatkov
