Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Threaded View
Hive, mail # user - hive.metastore.warehouse.dir - Should it point to a physical directory


Copy link to this message
-
Re: hive.metastore.warehouse.dir - Should it point to a physical directory
Sanjay Subramanian 2013-05-21, 18:27
Hi Raj

http://www.cloudera.com/content/cloudera-content/cloudera-docs/CDH4/latest/CDH4-Quick-Start/cdh4qs_topic_3.html

Installing CDH4 on a Single Linux Node in Pseudo-distributed Mode

On the left panel of the page u will find info on Hive installation etc.

I suggest CHD4 distribution only because it helps u to get started quickly…as developers I love to install from individual tar balls but sometimes there is little time to learn and execute

There are some great notes here

sanjay
From: bharath vissapragada <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>
Date: Tuesday, May 21, 2013 11:12 AM
To: "[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>" <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>, Raj Hadoop <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>
Cc: Sanjay Subramanian <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>, User <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>
Subject: Re: hive.metastore.warehouse.dir - Should it point to a physical directory
Yes !

On Tue, May 21, 2013 at 11:41 PM, Raj Hadoop <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>> wrote:
So that means I need to create a HDFS ( Not an OS physical directory ) directory under Hadoop that need to be used in the Hive config file for this property. Right?

From: Dean Wampler <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>
To: Raj Hadoop <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>
Cc: Sanjay Subramanian <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>; "[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>" <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>; User <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>
Sent: Tuesday, May 21, 2013 2:06 PM

Subject: Re: hive.metastore.warehouse.dir - Should it point to a physical directory

No, you only need a directory in HDFS, which will be "virtually located" somewhere in your cluster automatically by HDFS.

Also there's a typo in your hive.xml:

  <value/software/home/hadoop/hive/hive-0.9.0/warehouse</value>

Should be

  <value>/correct/path/in/hdfs/to/your/warehouse/directory</value>

On Tue, May 21, 2013 at 1:04 PM, Raj Hadoop <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>> wrote:
Thanks Sanjay.

My environment is  like this.

$ echo $HADOOP_HOME
/software/home/hadoop/hadoop/hadoop-1.1.2

$ echo $HIVE_HOME
/software/home/hadoop/hive/hive-0.9.0

$ id
uid=50052(hadoop) gid=600(apps) groups=600(apps)

So can i do like this:

$pwd
/software/home/hadoop/hive/hive-0.9.0

$mkdir warehouse

$cd /software/home/hadoop/hive/hive-0.9.0/warehouse

$ in hive-site.xml
<property>
  <name>hive.metastore.warehouse.dir</name>
  <value/software/home/hadoop/hive/hive-0.9.0/warehouse</value>
  <description>location of default database for the warehouse</description>
</property>

Where should I create the HDFS directory ?
From: Sanjay Subramanian <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>
To: "[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>" <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>; Raj Hadoop <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>; Dean Wampler <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>
Cc: User <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>
Sent: Tuesday, May 21, 2013 1:53 PM

Subject: Re: hive.metastore.warehouse.dir - Should it point to a physical directory

Notes below

From: Raj Hadoop <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>
Reply-To: "[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>" <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>, Raj Hadoop <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>
Date: Tuesday, May 21, 2013 10:49 AM
To: Dean Wampler <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>, "[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>" <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>
Cc: User <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>
Subject: Re: hive.metastore.warehouse.dir - Should it point to a physical directory

Ok.I got it. My questions -

1) Should a local physical directory be created before using this property?
I created a directory in HDFS during Hive installation
/user/hive/warehouse

My hive-site.xml has the following property defined

<property>
  <name>hive.metastore.warehouse.dir</name>
  <value>/user/hive/warehouse</value>
  <description>location of default database for the warehouse</description>
</property>

2) Should a HDFS file directory be created from Hadoop before using this property?
hdfs dfs -mkdir /user/hive/warehouse
Change the owner:group to hive:hive

From: Dean Wampler <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>
To: [EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>; Raj Hadoop <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>
Cc: User <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>
Sent: Tuesday, May 21, 2013 1:44 PM
Subject: Re: hive.metastore.warehouse.dir - Should it point to a physical directory

The name is misleading; this is the directory within HDFS where Hive stores the data, by default. (External tables can go elsewhere). It doesn't really have anything to do with the metastore.

dean

On Tue, May 21, 2013 at 12:42 PM, Raj Hadoop <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>> wrote:
Can some one help me on this ? I am stuck installing and configuring Hive with Oracle. Your timely help is really aprreciated.

From: Raj Hadoop <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>
To: Hive <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>; User <[EMAIL PROTECTED]<mailto:[EMAIL PROTECTED]>>
Sent: Tuesday, May 21, 2013 1:08 PM
Subject: hive.metastore.warehouse.dir - Should it point to a physical directory

Hi,

I am configurinig Hive. I ahve a question on the property hive.metastore.warehouse.dir.

Should this point to a physical directory. I am guessing it is a logical directory under Hadoop fs.default.name<http://fs.default.name/>. Please advise whether I