I'm migrating Java 1.7 to 1.8 and JbossEAP6.1 to WildFly14.0.1.
we are on AWS and we notice on the logs of the dockers that there are many errors of the type:
[GC (Allocation Failure) 1085489K->562034K(1908224K), 0.0661752 secs
and when we do performance tests (stress tests), sometimes dockers crashs.
we already analysed the heap dump and we didnt notice any thing special in our application code.
i'm using this configruation to start my WildFly Server :
JAVA_OPTS -Xms1920m -Xmx1920m -XX:MetaspaceSize=480m -XX:MaxMetaspaceSize=960m -XX:MaxPermSize=1920m -XX:+UseParallelGC -verbose:gc
can any one please help me on this subject.
Allocation failures are normal. They happen when a memory pool fills up. In your case, ParallelGC, they are what triggers a GC.
You had a GC pause of 66ms in which your heap went from 1'085'489K to 562'034K at a total size of 1'908'224K. You should see the same when you run the stress tests on EAP with Java 1.7 and PrallelGC.