Basically, I am trying to analyze SQL server log files using Hive. The layout is the log file is char(23) - for timestamp char(1) - space char(12) - source the rest of the row, and the length varies. The row delimiter is CR+LF. Below are some entries in the log. 2019-05-28 07:29:55.03 Server      UTC adjustment: -7:00 2019-05-28 07:29:55.03 Server      (c) Microsoft Corporation. 2019-05-28 07:29:55.03 Server      All rights reserved. 2019-05-28 07:29:55.03 Server      Server process ID is 3368. There are several posts here regarding fixed-width column layout. But in my case, the last column is identified not by the width but by the row delimiter. ... View more

Content of stock2.json file {"myid":"0001","mytype":"donut"} stock2.json file is in the path of '/warehouse/tablespace/external/hive/haijintest.db/stock2'; Below is the table creation script. CREATE EXTERNAL TABLE stock4_json (myjson struct<myid:string, mytype:string>)ROW FORMAT SERDE 'org.apache.hive.hcatalog.data.JsonSerDe' LOCATION '/warehouse/tablespace/external/hive/haijintest.db/stock2'; Below is the query that returned null value. 0: jdbc:hive2://> select myjson.myid from stock4_json; 19/05/27 20:20:18 [103d2343-eb59-4072-8a07-5469b2112093 main]: WARN optimizer.SimpleFetchOptimizer: Table haijintest@stock4_json is external table, falling back to filesystem scan. OK +-------+ | myid  | +-------+ | NULL  | +-------+ 1 row selected (0.232 seconds) 0: jdbc:hive2://> select * from stock4_json; OK +---------------------+ | stock4_json.myjson  | +---------------------+ | NULL                | +---------------------+ 1 row selected (0.242 seconds) Any suggestions? ... View more