[HIVE-14282] HCatLoader ToDate() exception with hive partition table ,partitioned by column of DATE datatype

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.2.1
• Fix Version/s: 1.3.0, 2.1.1, 2.2.0
• Component/s: HCatalog

Environment

PIG Version : (0.15.0)
HIVE : 1.2.1
OS Version : CentOS release 6.7 (Final)
OS Kernel : 2.6.32-573.18.1.el6.x86_64

• Hadoop Flags: Reviewed
• Assignee: Daniel Dai
• Reporter: Raghavender Rao Guruvannagari
• Created: 19/Jul/16 17:32
• Updated: 08/Dec/16 14:16
• Resolved: 20/Jul/16 22:20

Description

ToDate() function doesnt work with a partitioned table, partitioned by the column of DATE Datatype.
Below are the steps I followed to recreate the problem.
-->Sample input file to hive table :
hdfs@testhost ~$ cat test.log
2012-06-13,16:11:17,574,140.134.127.109,SearchPage,Google.com,Win8,5,HTC
2012-06-13,16:11:17,466,43.176.108.158,Electronics,Google.com,Win8,3,iPhone
2012-06-13,16:11:17,501,97.73.102.79,Appliances,Google.com,Android,4,iPhone
2012-06-13,16:11:17,469,166.98.157.122,Recommendations,Google.com,Win8,5,HTC
2012-06-13,16:11:17,557,36.159.147.50,Sporting,Google.com,Win8,3,Samsung
2012-06-13,16:11:17,449,128.215.122.234,ShoppingCart,Google.com,Win8,5,HTC
2012-06-13,16:11:17,502,46.81.131.92,Electronics,Google.com,Android,5,Samsung
2012-06-13,16:11:17,554,120.187.105.127,Automotive,Google.com,Win8,5,HTC
2012-06-13,16:11:17,447,127.94.64.59,DetailPage,Google.com,Win8,3,Samsung
2012-06-13,16:11:17,490,132.54.25.75,ShoppingCart,Google.com,Win8,3,iPhone
2012-06-13,16:11:17,578,79.201.53.179,Automotive,Google.com,Win8,5,Samsung
2012-06-13,16:11:17,435,158.106.164.38,HomePage,Google.com,Web,5,Chrome
2012-06-13,16:11:17,523,17.131.82.171,Recommendations,Google.com,Web,3,IE9
2012-06-13,16:11:17,575,178.95.126.105,Appliances,Google.com,iOS,3,iPhone
2012-06-13,16:11:17,468,225.143.39.176,SearchPage,Google.com,iOS,5,HTC
2012-06-13,16:11:17,511,43.103.102.147,ShoppingCart,Google.com,iOS,5,Samsung
--> Copied to hdfs directory:
hdfs@testhost ~$ hdfs dfs -put -f test.log /user/hdfs/
-->Create partitoned table (partitioned with date data type column) in hive:
0: jdbc:hive2://hdp2.raghav.com:10000/default> create table mytable(Dt DATE,Time STRING,Number INT,IPAddr STRING,Type STRING,Site STRING,OSType STRING,Visit INT,PhModel STRING) row format delimited fields terminated by ',' stored as textfile;
0: jdbc:hive2://testhost.com:10000/default> load data inpath '/user/hdfs/test.log' overwrite into table mytable;
0: jdbc:hive2://testhost..com:10000/default> SET hive.exec.dynamic.partition = true;
0: jdbc:hive2://testhost.com:10000/default> SET hive.exec.dynamic.partition.mode = nonstrict;
0: jdbc:hive2://testhost.com:10000/default> create table partmytable(Number INT,IPAddr STRING,Type STRING,Site STRING,OSType STRING,Visit INT,PhModel STRING) partitioned by (Dt DATE,Time STRING) row format delimited fields terminated by ',' stored as textfile;
0: jdbc:hive2://testhost.com:10000/default> insert overwrite table partmytable partition(Dt,Time) select Number,IPAddr,Type,Site,OSType,Visit,PhModel,Dt,Time from mytable;
0: jdbc:hive2://hdp2.raghav.com:10000/default> describe partmytable;
--> Try to filter with ToDate function which fails with error:
hdfs@testhost ~$ pig -useHCatalog
grunt>
grunt> temp = LOAD 'partmytable' using org.apache.hive.hcatalog.pig.HCatLoader();
grunt> temp1 = FILTER temp by dt == ToDate('2012-06-13','yyyy-MM-dd');
grunt> dump temp1;
-->Try to filter the normal table with same statement works;
grunt>
grunt> temp = LOAD 'mytable' using org.apache.hive.hcatalog.pig.HCatLoader();
grunt> temp1 = FILTER temp by dt == ToDate('2012-06-13','yyyy-MM-dd');
grunt> dump temp1;
Workaround :
Use below statement instead of direct ToDate();
grunt>temp1 = FILTER temp5 by DaysBetween(dt,(datetime)ToDate('2012-06-13', 'yyyy-MM-dd')) >=(long)0 AND DaysBetween(dt,(datetime)ToDate('2012-06-13', 'yyyy-MM-dd')) <=(long)0;