Normally you pass a directory as output, because each task is going to write a file. So wg.out seems to me like a directory.
2012/9/16 Sandy Ding <[email protected]> > yes, the output is supposed to be in hdfs /dataset/wg.out > when I 'ls' hdfs /dataset, wg.out exists but its size is 0 byte > and when I 'cat' the file, the shell complains that the file does not exist > > 2012/9/16 Thomas Jungblut <[email protected]> > > > If you run on a 3 node cluster, I assume you looked into HDFS right? > > > > 2012/9/16 Sandy Ding <[email protected]> > > > > > Hey, guys > > > > > > I run pagerank with the web-google input, it finished in36.554 > seconds. > > I > > > have a 3 node cluster and the job has 9 splits in total, is it normal? > (I > > > doubt it's too fast intuitively...) > > > The info from both the standord output and the log shows that the job > has > > > successfully finished and some statistics like SUPERSTEPS_SUM and > > > TOTOAL_MESSAGES_SENT was printed on the screen. But I find the output > > file > > > is 0 byte and I cannot 'cat' it. What might be wrong? > > > > > > Best > > > Sandy > > > > > >
