Hi Supun,
When I run twister2 with the local disk, it works with 10M tweetID-date pairs. It completed the membership finding job successfully with 240 workers. However, when I try to process 50M tweetID-date pairs with 240 workers. It gives an out of memory error.
I copied the java heap dump files of some workers to the following directory at the login node of victor:
/scratch_hdd/auyar/heap-dump/
$ ls
java_pid102168.hprof
java_pid102185.hprof
java_pid102208.hprof
java_pid102333.hprof
java_pid102444.hprof
logs
java_pid102172.hprof
java_pid102196.hprof
java_pid102272.hprof
java_pid102349.hprof
java_pid102475.hprof
java_pid102176.hprof
java_pid102200.hprof
java_pid102305.hprof
java_pid102353.hprof
java_pid102549.hprof
thanks,
Ahmet