Does HBaseStorage work with HBase 0.95?

 

This code was working with HBase 0.92 and Pig 0.9 but fails on HBase 0.95
and Pig 0.11 (built from source):

 

register /opt/hbase/hbase-trunk/hbase-0.95-SNAPSHOT.jar

register /opt/zookeeper/zookeeper-3.4.3/zookeeper-3.4.3.jar

 

 

tbl1 = LOAD 'input/sse.tbl1.HEADERLESS.csv' USING PigStorage( ',' ) AS (

      ID:chararray,

      hp:chararray,

      pf:chararray,

      gz:chararray,

      hid:chararray,

      hst:chararray,

      mgz:chararray,

      gg:chararray,

      epc:chararray );

 

STORE tbl1 INTO 'hbase://sse.tbl1' 

USING org.apache.pig.backend.hadoop.hbase.HBaseStorage('edrp:hp edrp:pf
edrp:gz edrp:hid edrp:hst edrp:mgz edrp:gg edrp:epc');

 

The job output (using either Grunt or PigServer makes no difference) shows
the family:descriptors being added by HBaseStorage then starts up the MR job
which (after a long pause) reports:

------------

Input(s):

Failed to read data from
"hdfs://namenode:8020/user/hadoop1/input/sse.tbl1.HEADERLESS.csv"

 

Output(s):

Failed to produce result in "hbase://sse.tbl1"

 

 

INFO mapReduceLayer.MapReduceLauncher: Failed!

INFO hbase.HBaseStorage: Adding family:descriptor filters with values
edrp:hp

INFO hbase.HBaseStorage: Adding family:descriptor filters with values
edrp:pf

INFO hbase.HBaseStorage: Adding family:descriptor filters with values
edrp:gz

INFO hbase.HBaseStorage: Adding family:descriptor filters with values
edrp:hid

INFO hbase.HBaseStorage: Adding family:descriptor filters with values
edrp:hst

INFO hbase.HBaseStorage: Adding family:descriptor filters with values
edrp:mgz

INFO hbase.HBaseStorage: Adding family:descriptor filters with values
edrp:gg

INFO hbase.HBaseStorage: Adding family:descriptor filters with values
edrp:epc

------------

 

The "Failed to read" is misleading I think because dump tbl1; in place of
the store works fine. 

 

I get nothing in the HBase logs and nothing in the Pig log.

 

HBase works fine from the shell and can read and write to the table. Pig
works fine in and out of HDFS on CSVs.

 

Any ideas?

 

Royston

 

Reply via email to