Try Setting up "udp" for the Kerberos Clients.   /etc/krb5.conf [libdefaults] udp_preference_limit = 1 ... View more

Hi w@leed    Thanks for Replying.   I did test the Job with all the three Collectors - ParallelGC, CMS and G1GC:   I has tested following options with the G1GC:   -XX:+UseG1GC -XX:+PrintGCDetails -XX:+PrintGCTimeStamps   and with CMS:   -XX:+UseConcMarkSweepGC -XX:+PrintGCTimeStamps -XX:CMSInitiatingOccupancyFraction=70 -XX:+UseParNewGC  -XX:+CMSConcurrentMTEnabled -XX:ParallelCMSThreads=10 -XX:ConcGCThreads=8 -XX:ParallelGCThreads=16 With G1GC defaults, I could see following: Desired survivor size 1041235968 bytes, new threshold 5 (max 15) [PSYoungGen: 1515304K->782022K(3053056K)] 2750361K->2017087K(6371840K), 1.5875321 secs] [Times: user=4.72 sys=0.74, real=1.59 secs] Heap after GC invocations=9 (full 3): PSYoungGen total 3053056K, used 782022K [0x0000000580000000, 0x000000068ef80000, 0x0000000800000000) eden space 2270720K, 0% used [0x0000000580000000,0x0000000580000000,0x000000060a980000) from space 782336K, 99% used [0x000000065f380000,0x000000068ef31ab0,0x000000068ef80000) to space 1016832K, 0% used [0x0000000612d80000,0x0000000612d80000,0x0000000650e80000) ParOldGen total 3318784K, used 1235064K [0x0000000080000000, 0x000000014a900000, 0x0000000580000000) object space 3318784K, 37% used [0x0000000080000000,0x00000000cb61e318,0x000000014a900000) Metaspace used 55055K, capacity 55638K, committed 55896K, reserved 1097728K class space used 7049K, capacity 7207K, committed 7256K, reserved 1048576K } {Heap before GC invocations=10 (full 3): PSYoungGen total 3053056K, used 3052742K [0x0000000580000000, 0x000000068ef80000, 0x0000000800000000) eden space 2270720K, 100% used [0x0000000580000000,0x000000060a980000,0x000000060a980000) from space 782336K, 99% used [0x000000065f380000,0x000000068ef31ab0,0x000000068ef80000) to space 1016832K, 0% used [0x0000000612d80000,0x0000000612d80000,0x0000000650e80000) ParOldGen total 3318784K, used 1235064K [0x0000000080000000, 0x000000014a900000, 0x0000000580000000) object space 3318784K, 37% used [0x0000000080000000,0x00000000cb61e318,0x000000014a900000) Metaspace used 55108K, capacity 55702K, committed 55896K, reserved 1097728K class space used 7049K, capacity 7207K, committed 7256K, reserved 1048576K 42.412: [GC (Allocation Failure) Desired survivor size 1653080064 bytes, new threshold 4 (max 15) [PSYoungGen: 3052742K->1016800K(3422720K)] 4287807K->2985385K(6741504K), 4.0304873 secs] [Times: user=11.87 sys=1.77, real=4.03 secs] Heap after GC invocations=10 (full 3): PSYoungGen total 3422720K, used 1016800K [0x0000000580000000, 0x0000000727a80000, 0x0000000800000000) eden space 2405888K, 0% used [0x0000000580000000,0x0000000580000000,0x0000000612d80000) from space 1016832K, 99% used [0x0000000612d80000,0x0000000650e78240,0x0000000650e80000) to space 1614336K, 0% used [0x00000006c5200000,0x00000006c5200000,0x0000000727a80000) ParOldGen total 3318784K, used 1968584K [0x0000000080000000, 0x000000014a900000, 0x0000000580000000) object space 3318784K, 59% used [0x0000000080000000,0x00000000f8272318,0x000000014a900000) Metaspace used 55108K, capacity 55702K, committed 55896K, reserved 1097728K class space used 7049K, capacity 7207K, committed 7256K, reserved 1048576K   With all the Collectors only difference I could see was that, a delayed full GC. I am considering to changing the YoungGen now. Will update if I do see a difference.   On a parallel note - 1. I did also see that there are some of the objects in the memory which remain persistent across GC cycles - for example : scala.Tuple2 and java.lang.Long  2. These are Java RDD's   Regards       ... View more