> to have at least 1 copy of at least X% of the blocks
Not 1, but dfs.replication.min, which by default is 1.
See also
http://wiki.apache.org/lucene-hadoop/FAQ#12
http://lucene.apache.org/hadoop/api/org/apache/hadoop/dfs/NameNode.html#setSafeMode(org.apache.hadoop.dfs.FSConstants.SafeModeAction)
--Konstantin
Owen O'Malley wrote:
On Oct 16, 2007, at 11:17 PM, cpreethi wrote:
It has been in safe mode all time. I am not bale to execute jobs
too. I find
some error sayin canot delete file. Namenode is in safemode.
Safe mode is the original state of the NameNode until it has a large
enough collection of DataNodes have reported in to have at least 1
copy of at least X% of the blocks. (The default X is 99.9%). I
recommend changing it to 100%. You can force the NameNode to leave
safe mode manually via "bin/hadoop dfsadmin -safemode leave". In safe
mode, the NameNode lets client do read-only operations but not write
operations (including replication decisions).
-- Owen
