On Thu, Mar 17, 2011 at 12:42 AM, Marcos Ortiz <mlor...@uci.cu> wrote: > 2011-03-15 01:18:44,468 INFO org.apache.hadoop.mapred.JobTracker: > problem cleaning system directory: > hdfs://localhost:9000/cygwin/usr/local/hadoop-datastore/hadoop-Baran/mapred/system > org.apache.hadoop.ipc.RemoteException: > org.apache.hadoop.hdfs.server.namenode.SafeModeException: Cannot > delete /cygwin/usr/local/hadoop-datastore/hadoop-Baran/mapred/system. > Name node is in safe mode.
Marcos, the JT keeps attempting to clear the mapred.system.dir on the DFS at startup, and fails because the NameNode wasn't ready when it tried (and thereby reattempts after a time, and passes later when NN is ready for some editing action). This is mostly because Baran is issuing a start-all/stop-all instead of a simple start/stop of mapred components. -- Harsh J http://harshj.com
