[
https://issues.apache.org/jira/browse/TRAFODION-199?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14696298#comment-14696298
]
Atanu Mishra commented on TRAFODION-199:
----------------------------------------
khaled Bouaziz (khaled-bouaziz) wrote on 2014-04-09: #2
The traces I added showed that that data is inserted in batched of 121 rows and
Esp 0 (there are 16 esps) is inserting the last batch of 121 rows twice before
it sees the EOD
The last batch in Esp 1 through 15 have are lees that 121 in size
I made some changes (see attached) and got the right results for the upsert
Orders query and seabase regression test suite
Is it possible to take a look at the changes and let me know
Weishiun Tsai (wei-shiun-tsai) on 2014-04-09
tags: added: sql-exe
removed: sql
Julie Thai (julie-y-thai) wrote on 2014-06-06: #3
Verified on RC2 (traf_20140606_0930):
Trafodion Conversational Interface 0.8.0
(c) Copyright 2014 Hewlett-Packard Development Company, LP.
>>create schema mysch;
--- SQL operation complete.
>>CREATE TABLE MYSCH.ORDERS
+> (
+> O_ORDERKEY INT NO DEFAULT NOT NULL
+> , O_CUSTKEY INT NO DEFAULT NOT NULL
+> , O_ORDERSTATUS CHAR(1) CHARACTER SET ISO88591 NO DEFAULT
NOT NULL
+>+> , O_TOTALPRICE NUMERIC(12, 2) NO DEFAULT NOT NULL
+> , O_ORDERDATE DATE NO DEFAULT NOT NULL
+> , O_ORDERPRIORITY CHAR(15) CHARACTER SET ISO88591 NO
+> DEFAULT NOT NULL
+> , O_CLERK CHAR(15) CHARACTER SET ISO88591 NO
+> DEFAULT NOT NULL
+> , O_SHIPPRIORITY INT NO DEFAULT NOT NULL
+> , O_COMMENT VARCHAR(79) CHARACTER SET ISO88591 NO
+> DEFAULT NOT NULL
+> , PRIMARY KEY (O_ORDERKEY ASC)
+> )
+>;
--- SQL operation complete.
>>upsert using load into MYSCH.ORDERS
+>select o_orderkey, o_custkey, o_orderstatus, o_totalprice,
+>cast(o_orderdate as date), o_orderpriority, o_clerk,
+>o_shippriority, o_comment from hive.hive.orders;
--- 3000000 row(s) inserted.
>>>>select count(*) from mysch.orders;
(EXPR)
--------------------
3000000
--- 1 row(s) selected.
Changed in trafodion:
status: Fix Committed → Fix Released
> LP Bug: 1294831 - Row count returned from upsert-using-load-into <traf-tbl>
> from <hive-tbl> may be incorrect.
> -------------------------------------------------------------------------------------------------------------
>
> Key: TRAFODION-199
> URL: https://issues.apache.org/jira/browse/TRAFODION-199
> Project: Apache Trafodion
> Issue Type: Bug
> Components: sql-exe
> Reporter: Julie Thai
> Assignee: Apache Trafodion
> Priority: Critical
> Fix For: 1.0 (pre-incubation)
>
>
> Symptom:
> Row count returned from 'upsert using load into...' is inconsistent with
> count(*) row count.
> >>upsert using load into mysch.orders select o_orderkey, o_custkey,
> >>o_orderstatus, o_totalprice, cast(o_orderdate as date), o_orderpriority,
> >>o_clerk, o_shippriority, o_comment from hive.hive.orders;
> --- 3000121 row(s) inserted.
> >>select count(*) from mysch.orders;
> (EXPR)
> --------------------
> 3000000
> --- 1 row(s) selected.
> To reproduce:
> 1. Setup hive table and load from text file (see attachment) with 3M rows:
> CREATE TABLE orders
> (
> O_ORDERKEY INT
> , O_CUSTKEY INT
> , O_ORDERSTATUS STRING
> , O_TOTALPRICE DOUBLE
> , O_ORDERDATE TIMESTAMP
> , O_ORDERPRIORITY STRING
> , O_CLERK STRING
> , O_SHIPPRIORITY INT
> , O_COMMENT STRING
> )
> row format delimited fields terminated by '|'
> ;
> load data local inpath './orders.tbl' into table orders;
> 2. In traf, create table and upsert-using-load from hive table:
> create schema MYSCH;
> CREATE TABLE MYSCH.ORDERS
> (
> O_ORDERKEY INT NO DEFAULT NOT NULL
> , O_CUSTKEY INT NO DEFAULT NOT NULL
> , O_ORDERSTATUS CHAR(1) CHARACTER SET ISO88591 NO
> DEFAULT
> NOT NULL
> , O_TOTALPRICE NUMERIC(12, 2) NO DEFAULT NOT NULL
> , O_ORDERDATE DATE NO DEFAULT NOT NULL
> , O_ORDERPRIORITY CHAR(15) CHARACTER SET ISO88591 NO
> DEFAULT NOT NULL
> , O_CLERK CHAR(15) CHARACTER SET ISO88591 NO
> DEFAULT NOT NULL
> , O_SHIPPRIORITY INT NO DEFAULT NOT NULL
> , O_COMMENT VARCHAR(79) CHARACTER SET ISO88591 NO
> DEFAULT NOT NULL
> , PRIMARY KEY (O_ORDERKEY ASC)
> )
> ;
> upsert using load into MYSCH.ORDERS
> select o_orderkey, o_custkey, o_orderstatus, o_totalprice,
> cast(o_orderdate as date), o_orderpriority, o_clerk,
> o_shippriority, o_comment from hive.hive.orders;
> select count(*) from MYSCH.ORDERS;
> Assigned to LaunchPad User khaled Bouaziz
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
