AnsweredAssumed Answered

Partition creation... no more data to read.

Question asked by toadrw on Mar 6, 2013
Latest reply on Mar 13, 2013 by peterconrad

Hello all:

I know this table is set up wrong to add a partition to it.  However, I ran the SQL anyways. :p  The table creates, but the next statement fails and causes all subsequent queries to fail. :/

I'm finding that when I run the alter table statement below I get a "no more data to read" error which is fine as I'm expecting an error, but after that I can't run any queries via ODBC.  They all seem to fail with error 10061 and I need to reboot the machine for everything to start working again.  Not sure if this is a Hadoop issue or ODBC issue, but I thought I should report it to see if you can reproduce.

I'm running Hive .10 on Cloudera's virtual machine.  I'm using the MapR .8 ODBC driver (32-bit) so that may be the issue as the versions are different, but hopefully this is something which can be resolved in the .10 version of the driver (if it is not a Hadoop issue).  Thank you for looking at this.

CREATE TABLE IF NOT EXISTS sales_table
(Product_ID int,
Sale_timestamp  timestamp,
Daily_Sales  string)
ROW FORMAT DELIMITED
FIELDS TERMINATED BY ','
STORED AS TEXTFILE;

ALTER TABLE sales_table ADD IF NOT EXISTS PARTITION (Daily_Sales='2000-09-28', Product_Id=3000)
PARTITION (Daily_Sales='2000-09-29', Product_Id=1000) PARTITION (Daily_Sales='2000-09-27', Product_Id = 2000);

Outcomes