Hi all,
I got this exception when ran my job. Then, I ran the PageRank in
hama-example and found that same data in different order leads to
different result . I modified the vertex reader of PageRank.java and
setInputFormat(TextInputFormat.class) to process a text file, while
other code of PageRank.java remain unchanged .
If the input file :
3\t2\t1<br/>
2\t1<br/>
1<br/>
PageRank runs well .
If I deleted the last line (which represents a single page "1") :
3\t2\t1<br/>
2\t1<br/>
Job failed and I got the exception "Messages must never be behind the
vertex in ID! Current Message ID: 1 vs. 2"
What's more, If the input file contains only one line :
3\t2\t1<br/>
Job failed with the exception "Messages must never be behind the vertex
in ID! Current Message ID: 1 vs. 3".
But when I exchanged the situations of "3" and "1" :
1\t2\t3<br/>
It ran well.
What's wrong ? Do I overlook something or It is a bug ?
Regards,
Lyu.
|