[HIVE-4969] HCatalog HBaseHCatStorageHandler is not returning all the data

• Type: Bug
• Status: Open
• Priority: Critical
• Resolution: Unresolved
• Affects Version/s: 0.11.0
• Component/s: HCatalog
• Assignee: Venki Korukanti
• Reporter: Venki Korukanti
• Created: 31/Jul/13 22:49
• Updated: 27/Oct/13 17:13

Description

Repro steps:
1) Create an HCatalog table mapped to HBase table.

hcat -e "CREATE TABLE studentHCat(rownum int, name string, age int, gpa float)
STORED BY 'org.apache.hcatalog.hbase.HBaseHCatStorageHandler'
TBLPROPERTIES('hbase.table.name' ='studentHBase',
'hbase.columns.mapping' =
':key,onecf:name,twocf:age,threecf:gpa')";

2) Load the following data from Pig.

cat student_data
1^Asarah laertes^A23^A2.40
2^Atom allen^A72^A1.57
3^Abob ovid^A61^A2.67
4^Aethan nixon^A38^A2.15
5^Acalvin robinson^A28^A2.53
6^Airene ovid^A65^A2.56
7^Ayuri garcia^A36^A1.65
8^Acalvin nixon^A41^A1.04
9^Ajessica davidson^A48^A2.11
10^Akatie king^A39^A1.05

grunt> A = LOAD 'student_data' AS (rownum:int,name:chararray,age:int,gpa:float);

grunt> STORE A INTO 'studentHCat' USING org.apache.hcatalog.pig.HCatStorer();

3) Now from HBase do a scan on the studentHBase table
hbase(main):026:0> scan 'studentPig',

{LIMIT => 5}

4) From pig access the data in table
grunt> A = LOAD 'studentHCat' USING org.apache.hcatalog.pig.HCatLoader();
grunt> STORE A INTO '/user/root/studentPig';

5) Verify the output written in StudentPig
hadoop fs -cat /user/root/studentPig/part-r-00000
1 23
2 72
3 61
4 38
5 28
6 65
7 36
8 41
9 48
10 39

The data returned has only two fields (rownum and age).

Problem:
While reading the data from HBase table, HbaseSnapshotRecordReader gets data row in Result (org.apache.hadoop.hbase.client.Result) object and processes the KeyValue fields in it. After processing, it creates another Result object out of the processed KeyValue array. Problem here is KeyValue array is not sorted. Result object expects the input KeyValue array to have sorted elements. When we call the Result.getValue() it returns no value for some of the fields as it does a binary search on un-ordered array.