What is the difference between hadoop fs put and copyFromLocal?

The answer to this question depends on the version of Hadoop you are using.

Older version of Hadoop (1.x.x)

The first argument of copyFromLocal is restricted to a location in the location filesystem whereas  you are not restricted to the local file system with the put command.

With put you can specify the filesystem scheme (file:// or hdfs://) to distinguish between local filesystem and HDFS.

copyFromLocal

hdfs dfs -copyFromLocal <local source> <destination>

put

hdfs dfs -put <source> <destination>

Newer versions of Hadoop (> 2.0.0)

With the newer versions of Hadoop, put and copyFromLocal does exactly the same. Infact copyFromLocal calls the -put command. You can see this by calling the help on the commands.

[hirw@wk1 ~]$ hdfs dfs -help put

-put [-f] [-p] [-l] <localsrc> … <dst> :

  Copy files from the local file system into fs. Copying fails if the file already

  exists, unless the -f flag is given.

  Flags:

  -p  Preserves access and modification times, ownership and the mode.

  -f  Overwrites the destination if it already exists.

  -l  Allow DataNode to lazily persist the file to disk. Forces

         replication factor of 1. This flag will result in reduced

         durability. Use with care.

[hirw@wk1 ~]$ hdfs dfs -help copyFromLocal

-copyFromLocal [-f] [-p] [-l] <localsrc> … <dst> :

  Identical to the -put command.

[hirw@wk1 ~]$

From the code you can see the copyFromLocal extends put.

public static class CopyFromLocal extends Put { 
  public static final String NAME = "copyFromLocal"; 
  public static final String USAGE = Put.USAGE; 
  public static final String DESCRIPTION = "Identical to the -put command."; 
}