hadoop hive

Apache Hive 3.0.0 Installation on Windows 10 Step by Step Guide

7,195 views 9 comments about 7 months ago Raymond Tang

If you have been following my website, you would know I’ve published a number of articles about installing big data tools/frameworks on Windows. You probably will ask, what reasons are to install these tools on Windows? Well, if you don’t have a Hadoop platform, it will be handy to have them installed on your Windows so that you can play around for learning and practices.

On this page, I’m going to demo how to install Hive 3.0.0 on Windows 10.

Prerequisites

Install Hadoop 3.0.0

Follow this page to install Hadoop 3.0.0 first.

Install Hadoop 3.0.0 in Windows (Single Node)

Tools and Environment

  • Windows 10
  • Cygwin
  • Command Prompt

Install Cygwin

Please install Cygwin so that we can run Linux shell scripts on Windows. From Hive 2.3.0, the binary doesn’t include any CMD file anymore. Thus you have to use Cygwin or any other similar tools to run the scripts.

You can install Cygwin from this site: https://www.cygwin.com/.

Download Binary Package

Download the latest binary from the official website:

https://hive.apache.org/downloads.html

Save the downloaded package to a local drive. For my case, I am saving to ‘F:\DataAnalytics’.

If you cannot find the package, you can download from the archive site too: https://archive.apache.org/dist/hive/hive-3.0.0/.

UnZip binary package

Open Cygwin terminal, and change directory (cd) to the folder where you save the binary package and then unzip:

$ cd F:\DataAnalytics

fahao@Raymond-Alienware /cygdrive/f/DataAnalytics
$ tar -xvzf apache-hive-3.0.0-bin.tar.gz

Setup environment variables

Run the following commands in Cygwin to setup the environment variables:

export HADOOP_HOME='/cygdrive/f/DataAnalytics/hadoop-3.0.0'
export PATH=$PATH:$HADOOP_HOME/bin
export HIVE_HOME='/cygdrive/f/DataAnalytics/apache-hive-3.0.0-bin'
export PATH=$PATH:$HIVE_HOME/bin
export HADOOP_CLASSPATH=$HADOOP_CLASSPATH:$HIVE_HOME/lib/*.jar

You can add these exports to file .bashrc so that you don’t need to run these command manually each time when you launch Cygwin:

vi ~/.bashrc

Setup Hive HDFS folders

Open Command Prompt (not Cygwin) and then run the following commands:

hadoop fs -mkdir /tmp

hadoop fs -mkdir -p /user/hive/warehouse

hadoop fs -chmod g+w /tmp

hadoop fs -chmod g+w /user/hive/warehouse

Create a symbolic link

As Java doesn’t understand Cygwin path properly, you may encounter errors like the following:

JAR does not exist or is not a normal file: F:\cygdrive\f\DataAnalytics\apache-hive-3.0.0-bin\lib\hive-beeline-3.0.0.jar

In my system, Hive is installed in F:\DataAnalytics\ folder. To make it work, follow these steps:

  • Create a folder in F: driver named cygdrive
  • Open Command Prompt (Run as Administrator) and then run the following command:

C:\WINDOWS\system32>mklink /J  F:\cygdrive\f\ F:\
Junction created for F:\cygdrive\f\ <<===>> F:\

In this way, ‘F:\cygdrive\f’ will be equal to ‘F:\’.  You need to change the drive to the appropriate drive where you are installing Hive. For example, if you are installing Hive in C driver, the command line will be:

C:\WINDOWS\system32>mklink /J  C:\cygdrive\c\ C:\

Initialize metastore

Now we need to initialize the schemas for metastore.

$HIVE_HOME/bin/schematool -dbType <db type> -initSchema

Type the following command to view all the options:

$HIVE_HOME/bin/schematool -help

For argument dbType, the value can be one of the following databases:

derby|mysql|postgres|oracle|mssql

In my system, I am going to use derby as it is purely Java based and also already included in the Hive release:

$HIVE_HOME/bin/schematool -dbType derby -initSchema

The output looks similar to the following:

Metastore connection URL:        jdbc:derby:;databaseName=metastore_db;create=true
Metastore Connection Driver :    org.apache.derby.jdbc.EmbeddedDriver
Metastore connection User:       APP
Starting metastore schema initialization to 3.0.0
Initialization script hive-schema-3.0.0.derby.sql

Initialization script completed
schemaTool completed

A folder named metastore_db will be created on your current path (pwd).

Configure a remote database as metastore

To support multiple sessions, please refer to this post about configuring SQL Server database as metastore.

Configure a SQL Server Database as Remote Hive Metastore

* I would recommend using a remote database as metastore for Hive.

Configure API authentication

Add the following configuration into hive-site.xml file.

<property>
    <name>hive.metastore.event.db.notification.api.auth</name>
     <value>false</value>
     <description>
       Should metastore do authorization against database notification related APIs such as get_next_notification.
       If set to true, then only the superusers in proxy settings have the permission
     </description>
   </property>

Or configure proxy user in Hadoop core-site.xml file. Refer to the following post for more details:

HiveServer2 Cannot Connect to Hive Metastore Resolutions/Workarounds

Start HiveServer2 service

$HIVE_HOME/bin/hive --service hiveserver2 start

Start HiveServer2 service and run Beeline CLI

Now you can run the following command to start HiveServer2 service and Beeline in the same process:

$HIVE_HOME/bin/beeline -u jdbc:hive2://
image

Run CLI directly

You can also  run the CLI either via hive or beeline command.

$HIVE_HOME/bin/beeline

$HIVE_HOME/bin/hive

image

image

You can also specify beeline commands with JDBC URL of HiveServer2.  In the following command, you need to replace $HS2_HOST with HiveServer2 address and $HS2_PORT with HiveServer2 port.

$HIVE_HOME/bin/beeline -u jdbc:hive2://$HS2_HOST:$HS2_PORT

By default the URL is: jdbc:hive2://localhost:10000.

Until now, we have installed the clients successfully.

DDL practices

Now we have Hive installed successfully, we can run some commands to test.

For more details about the commands, refer to official website:

https://cwiki.apache.org/confluence/display/Hive/HiveServer2+Clients

Create a new Hive database

Run the following command in Beeline to create a database named test_db:

create database if not exists test_db;

Output of the command looks similar to the following:

0: jdbc:hive2://> create database if not exists test_db;
19/03/26 21:44:09 [HiveServer2-Background-Pool: Thread-115]: WARN metastore.ObjectStore: Failed to get database hive.test_db, returning NoSuchObjectException
OK
No rows affected (0.312 seconds)

As I didn’t specify the database location, it will be created under the default HDFS location: /user/hive/warehouse.

In HDFS name node, we can see a new folder is created.

image

Create a new Hive table

Run the following commands to create a table named test_table:

use test_db;

create table test_table (id bigint not null, value varchar(100));

show tables;

Insert data into Hive table

Run the following command to insert some sample data:

insert into test_table (id,value) values (1,'ABC'),(2,'DEF');

Two records will be created by the above command.

The command will submit a MapReduce job to YARN. You can also configure Hive to use Spark as execution engine instead of MapReduce.

You can track the job status through Tracking URL printed out by the console output.

Go to YARN, you can also view the job status:

image

Wait until the job is completed.

Select data from Hive table

Now, you can display the data by running the following command in Beeline:

select * from test_table;

The output looks similar to the following:

0: jdbc:hive2://> select * from test_table;
19/03/26 23:23:18 [93fd08aa-09f6-488a-aa43-28b37d69a504 main]: WARN metastore.ObjectStore: datanucleus.autoStartMechanismMode is set to unsupported value null . Setting it to value: ignored
OK
+----------------+-------------------+
| test_table.id  | test_table.value  |
+----------------+-------------------+
| 1              | ABC               |
| 2              | DEF               |
+----------------+-------------------+
2 rows selected (0.476 seconds)

In Hadoop NameNode website, you can also find the new files are created:

image

Now you can enjoy working with Hive on Windows!

Fix some errors

hadoop-3.0.0/bin/hadoop: line 2: $'\r': command not found

I got this error in my environment because end line is in Windows format instead of UNIX.

/cygdrive/f/DataAnalytics/apache-hive-3.1.1-bin
$ $HIVE_HOME/bin/beeline
F:\DataAnalytics\hadoop-3.0.0/bin/hadoop: line 2: $'\r': command not found
F:\DataAnalytics\hadoop-3.0.0/bin/hadoop: line 17: $'\r': command not found
F:\DataAnalytics\hadoop-3.0.0/bin/hadoop: line 20: $'\r': command not found
F:\DataAnalytics\hadoop-3.0.0/bin/hadoop: line 26: syntax error near unexpected token `$'{\r''
':\DataAnalytics\hadoop-3.0.0/bin/hadoop: line 26: `{
Unable to determine Hadoop version information.
'hadoop version' returned:
F:\DataAnalytics\hadoop-3.0.0/bin/hadoop: line 2: $'\r': command not found
F:\DataAnalytics\hadoop-3.0.0/bin/hadoop: line 17: $'\r': command not found
F:\DataAnalytics\hadoop-3.0.0/bin/hadoop: line 20: $'\r': command not found
F:\DataAnalytics\hadoop-3.0.0/bin/hadoop: line 26: syntax error near unexpected token `$'{\r''
':\DataAnalytics\hadoop-3.0.0/bin/hadoop: line 26: `{

image

This can be confirmed by opening the hadoop file in Notepad++.

To fix this, click Edit -> EOL Conversion -> Unix (LF).

Do the same for all the other shell scripts if similar errors occur.

I applied the same fix to my following scripts in folder $HADOOP_HOME/bin:

  • hdfs
  • mapred
  • yarn

If you encounter any other issues, feel free to post a comment here. I will try to help as much as I can.

Add comment

Comments (8)

Raym*** about 13 hours ago

Did you get Hadoop installation successfully first?

This is a HDFS CLI issue and is not related to Hive installation.

Swat*** about 1 day ago

Hi Team,

At step, Set up Hive HDFS folders while creating dir using hadoop fs -mkdir /tmp at cmd, the system is throwing an error.

mkdir: Your endpoint configuration is wrong

Please suggest how to resolve this.

Regards,

Swati

Swat*** about 1 day ago

Hi Team,

At step, Set up Hive HDFS folders while creating dir using hadoop fs -mkdir /tmp at cmd, the system is throwing an error.

mkdir: Your endpoint configuration is wrong

Please suggest how to resolve this.

Regards,

Swati

Raym*** about 6 months ago

You mentioned $HIVE_PATH in one of the previous comments while it should be $HIVE_HOME. Can you please double check that?

Based on what you have described and also if I understand correctly:

Your issue is that you cannot run the following command successfully:

$HIVE_HOME/bin/schematool -dbType derby -initSchema

And you got error: no such file or directory exist.

Usually this issue will happen if:

  • No x permission (execute) for your account on schematool file, which is why I recommended to check that permission and add it if missing.
  • Or as the error message self-described, the path doesn't exist. For example, it may be because your $HIVE_HOME environment variable is not setup correctly. I would recommend to follow the steps below to add it into your .bashrc file and then re-run the schema initialisation command: 

1) Edit file ~/.bashrc by running the following command

vi ~/.bashrc

2) Add the following line at the end of the file:

export HIVE_HOME={your hive home folder path}

3) Source the settings 

source ~/.bashrc

And then run the command:

$HIVE_HOME/bin/schematool -dbType derby -initSchema

If the above suggestions don't work still, I'm not sure whether I can help more unless you provide screenshots about your Cygwin window, hive folder,  and detailed error messages here.

You can upload images in the comment section directly.  Or alternatively, write me an email at enquiry[at]kontext.tech

Ba*** about 6 months ago

So, $HIVE_HOME/bin was in the path. So, I just ran schematool -dbType mysql -initSchema. Also, before that I did hive -service metastore.

Ba*** about 6 months ago

So, $HIVE_HOME/bin was in the path. So, I just ran schematool -dbType mysql -initSchema. Also, before that I did hive -service metastore.

Raym*** about 6 months ago

Can you run the following command in Cygwin to see if the script file is executable?

ls -alt $HIVE_HOME/bin

-rwxr-xr-x+ 1 fahao fahao   832 May 16  2018 metatool

-rwxr-xr-x+ 1 fahao fahao   884 May 16  2018 schematool

The output should look like the above one. 'x' means execution permission. You will need that permission before you can execute the scripts. 

If no permission, please try the following command to add it:

chmod +x $HIVE_HOME/bin/schematool

Ba*** about 6 months ago

Yes, I did. When I go the path from cygwin, and do a ls, I see schematool there. Also when I print $HIVE_PATH and $HADOOP_PATH I get the correct location.

Ba*** about 6 months ago

Yes, I did. When I go the path from cygwin, and do a ls, I see schematool there. Also when I print $HIVE_PATH and $HADOOP_PATH I get the correct location.

Raym*** about 6 months ago

Did you run the command in Cygwin terminal? $HIVE_HOME is the syntax for Linux/UNIX and only works in Cygwin (or other equivalent terminal) in Windows.

Ba*** about 6 months ago

When I try to run $HIVE_HOME/bin/schematool -dbType derby -initSchema I get no such file or directory exist. But when I go to the exact location with cd and run ls, I see the file there. Also, when I echo HIVE_HOME it return me the exact path.

Ba*** about 6 months ago

When I try to run $HIVE_HOME/bin/schematool -dbType derby -initSchema I get no such file or directory exist. But when I go to the exact location with cd and run ls, I see the file there. Also, when I echo HIVE_HOME it return me the exact path.