"Counters from Step 1: No Counters found" using Hadoop and mrjob
One problem is in the encoding of the bigram of the mapper. The way it is coded above, bigram is the python type "tuple":
>>> word = 'the'>>> word2 = 'boy'>>> bigram = word, word2>>> type(bigram)<type 'tuple'>
Usually, plain strings are used as the keys. So instead, create bigram as a string. One way you could do that is:
bigram = '-'.join((word, nextword))
When I make that change in your program, then I see outputs like this:
automatic-translation 1automatic-vs 1automatically-focus 1automatically-learn 1automatically-learning 1automatically-translate 1available-including 1available-without 1
One other hint: try -q
on your command line to silence all of the hadoop intermediate noise. Sometimes it just gets in the way.
HTH.
This is a cache error . I mostly found this with Hortonworks sandbox . Simple solution is to logout from the sandbox and ssh again ..