We are trying to run Hadoop version 1.2.1 single node cluster on a 4 core Intel server. We are trying to run the hadoop programs only on 3 cores instead of 4. We have tried using taskset but it didn't work. We tried using cset as well but it was not good. Our theory was the init which spawns the hadoop processes (Namenode, Datanode, etc) make their affinity to all cores, so we couldn't manually change their affinity using taskset. Is there any option on hadoop-env that we can use? I couldn't find any documentation online. Any help would be appreciated.
Asked
Active
Viewed 37 times
