Good morning. After my local Hadoop User Group meeting last night, I decided to switch over from using the native "hive" shell to "beeline." I can't remember the exact reasons, but the wonderful speaker made a point of saying users needed to back way from using the native "hive" shell for some very good reasons that I've forgotten after three beers last night.
In anysense, I took the advice and fired up beeline this morning. Everything seems to be working well, but when trying to load data, I get an "Invalid Path" error. Below you can see that when not fully qualifying the file name, the working directory is set to where the properties and such are stored. That's fine.
Connected to: Apache Hive (version 0.13.1-cdh5.2.0)
Driver: Hive JDBC (version 0.13.1-cdh5.2.0)
Transaction isolation: TRANSACTION_REPEATABLE_READ
Beeline version 0.13.1-cdh5.2.0 by Apache Hive
0: jdbc:hive2://c-10-206-76-8.int.cis.trcloud> load data local inpath 'hourly_TEMP_2014.csv' into table temps_txt;
Error: Error while compiling statement: FAILED: SemanticException Line 1:23 Invalid path ''hourly_TEMP_2014.csv'': No files matching path file:/var/run/cloudera-scm-agent/process/24-hive-HIVESERVER2/hourly_TEMP_2014.csv (state=42000,code=40000)
This is fine, I can fully qualify the file. However, even when I do that, I still get that the file is not found.
0: jdbc:hive2://c-10-206-76-8.int.cis.trcloud> load data local inpath '/home/cloud/hourly_TEMP_2014.csv' into table temps_txt;
Error: Error while compiling statement: FAILED: SemanticException Line 1:23 Invalid path ''/home/cloud/hourly_TEMP_2014.csv'': No files matching path file:/home/cloud/hourly_TEMP_2014.csv (state=42000,code=40000)
0: jdbc:hive2://c-10-206-76-8.int.cis.trcloud> !quit
Closing: 0: jdbc:hive2://c-10-206-76-8.int.cis.trcloud:10000/default
cloud@c-10-206-76-8:~> ls -l /home/cloud/* | grep TEMP
-rw-rw-r-- 1 cloud cloud 1180101268 Jan 22 21:28 /home/cloud/hourly_TEMP_2014.csv
When I issue these commands via the "hive" shell, the file location resolves fine - both relatively and fully qualified. I'm going to upgrade my small cluster to CDH 5.3.0 to see if the Hive version + backports change the behavior, but figured I'd post this to see if anyone has seen this issue with the 5.2.0 release.
Thanks for your time.
(Oh, also in the "Labels" section of this forum, there is only CDH 4.x options to choose from and it's a required field. So I selected 4.6.x, even though this is realted to CDH 5.2.0. Just thought I'd note that. Or it could be that I'm in the wrong area. Wouldn't be the first time.)