Hi, I'm using two hadoop clusters, 0.21.0 and CDH2. The problem which resembled MAPREDUCE-1868 by my clusters by hanging up of TaskTracker occurred.
Is it having to take the following methods in order to solve this problem? 1. The patch contributed by MAPREDUCE-1868 and the similar patch are applied to 0.21.0 and CDH2. 2. Property "mapreduce.client.output.filter" is changed into "NONE" (However, FAILED-tasklog disappears from JobClient). Regards, Shinichi YAMASHITA
