You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@hive.apache.org by "Alexandre Linte (JIRA)" <ji...@apache.org> on 2017/09/29 17:06:00 UTC

[jira] [Created] (HIVE-17654) Beeline dfs -get retrieved file on Hiveserver2 server at /home/hiveserver2 location.

Alexandre Linte created HIVE-17654:
--------------------------------------

             Summary: Beeline dfs -get retrieved file on Hiveserver2 server at /home/hiveserver2 location.
                 Key: HIVE-17654
                 URL: https://issues.apache.org/jira/browse/HIVE-17654
             Project: Hive
          Issue Type: Bug
          Components: Beeline
    Affects Versions: 2.1.0
         Environment: Centos6.8. VM. Remote Desktop connected to hiveserver2.
            Reporter: Alexandre Linte


Hello,

An user tried to retrieved an HDFS files from Beeline to its local /home/{user} repository. For that he used the `dfs -get /hdfs/path/to/file.csv file.csv` command from beeline client.
First try, it seems to work. The ouput is the following:
+-------------+--+
  DFS Output |
+-------------+--+
+-------------+--+

Second and furthur tries lead to the following error:
Error: Error while processing statement: null (state=,code=1)

It seems that the error was due to the file already existing in the current folder. However he couldn't find the file in its current folder, the VM on which he is running beeline client connected to the hiveserver2 server.

After looking into the matter, we found its file (files because he did several tries) on the hiveserver2 server at /home/hiveserver2 location.

Is it the expected behavior ? When using Hive client, The file is retrieved as expected on the current directory and not on the hiveserver2 under /home/hiveserver2 location. In fact the user change from Hive client to beeline as Hive client is considered as deprecated. He was expecting to get the same result from beeline.
After looking a bit into our Hive configuration file, we did not found a parameter to change this beeline behavior. Did we missed something ? Or is there a way to change this behavior in order for beeline to work the same way as Hive client for dfs commands ?

Also we are wondering if it could lead to a "security breach". What we mean is anyone that have the access to an hiveserver2 server through beeline could blow it up by retrieving files until "no space left on device".

Regards.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)