Hi Yongcheng, As far as I know, currently there is no easy way of achieving. We are working on this feature based on HCat/Hive. Dev works progress can be found at: https://issues.apache.org/jira/browse/OOZIE-561
At this time, I can think of writing a Hive action or Java action to get those range of instance from metastore. I know it is not convenient. Regards, Mohammad ________________________________ From: Yongcheng Li <[email protected]> To: "[email protected]" <[email protected]> Sent: Wednesday, November 14, 2012 11:44 AM Subject: Synchronous Datasets for Hive Hi, Synchronous datasets provides clocked datasets for map-reduce coordinator applications. However, I am not clear how to use it in my Hive action. I have a Hive script that processes the past seven days' data. What I need to do is either [1] Empty the table content and then using the following statement to load the desired data into Hive table and then run my Hive script to process the data. Day1: ALTER TABLE myTable ADD IF NOT EXISTS PARTITION (dt = 'mm-dd-yyyy') LOCATION '/input/yyyy/mm/dd'; Day2: ... .. ... Day7: ... ... Or []2 Load the daily partition into Hive table and use Hive partition in my select statement. Anyway, I need to access dynamic YEAR, MONTH, DAY values (similar to the way "start-instance" and "end-instance" are used) generated by my coordinator application in my Hive script. Any suggestion on what is possibly a good approach? Thanks! Yongcheng
