Uninstalling on a Cluster

Uninstalling the Hadoop Distribution

These instructions assume that you already have a Hadoop distribution installed on your system and, in the case of Cloudera and Hortonworks, that Trafodion is also already installed. They explain how to uninstall the Hadoop distribution using the cloudera_uninstall, hortonworks_uninstall, or mapr_uninstall script so that you can then install a different, supported Hadoop distribution to be used with Trafodion.

• Uninstalling Cloudera
• Uninstalling Hortonworks/Ambari
• Uninstalling MapR

Uninstalling Cloudera

The following commands should all be done from the first node of the cluster. Do not run them from a machine that is not part of the Trafodion cluster. Follow these steps only if you are uninstalling Cloudera and not uninstalling Trafodion.

1. Stop Trafodion:

   su trafodion
   cd $MY_SQROOT/sql/scripts or cds
   sqstop
   

2. Run cloudera_uninstall:

   NOTE: You must have root access to the machine.

   Required Parameters	Description
   --help	Print this message and exit.
   --nodes "<node_list>"	Space-separated list of nodes (within quotes)

   EXAMPLES:
   cd /root/trafodion_downloads/installer/tools
   ./cloudera_uninstall --nodes "redhat-1 redhat-2"
   

3. Reinstall a supported Hadoop distribution. See Installing a Supported Hadoop Distribution.
4. Restart Trafodion:

   su trafodion
   cd $MY_SQROOT/scripts or cds
   sqstart
   

5. Perform a quick sanity check by creating a table with a couple of records using the Trafodion Conversational Interface (sqlci). For example:

   [trafodion@n001 ~]$ sqlci
   Trafodion Conversational Interface 1.1.0
   (c) Copyright 2015 Hewlett-Packard Development Company, LP.
   >>create table test1 (f1 int, f2 int);
   
   --- SQL operation complete.
   >>insert into test1 values(1,1);
   
   --- 1 row(s) inserted.
   >>insert into test1 values(2,2);
   
   --- 1 row(s) inserted.
   >>select * from test1;
   
   F1            F2
   -----------   -----------
             1             1
             2             2
   
   --- 2 row(s) selected.
   >>get tables;
   
   Tables in Schema TRAFODION.SEABASE
   ==================================
   
   TEST1
   
   --- SQL operation complete.
   >>exit;
   

   Those SQL commands should run successfully.

Uninstalling Hortonworks/Ambari

The following commands should all be done from the first node of the cluster. Do not run them from a machine that is not part of the Trafodion cluster. Follow these steps only if you are uninstalling Hortonworks/Ambari and not uninstalling Trafodion.

1. Stop Trafodion:

   su trafodion
   cd $MY_SQROOT/sql/scripts or cds
   sqstop
   

2. Run hortonworks_uninstall:

   NOTE: You must have root access to the machine.

   Required Parameters	Description
   --help	Print this message and exit.
   --nodes "<node_list>"	Space-separated list of nodes (within quotes)

   EXAMPLES:
   cd /root/trafodion_downloads/installer/tools
   ./hortonworks_uninstall --nodes "redhat-1 redhat-2"
   

3. During uninstall, the Ambari Server database must be reset. Type 'yes' when prompted.

   EXAMPLE:
   Resetting ambari-server
   **** WARNING **** You are about to reset and clear the Ambari Server database. This will remove all cluster host and configuration 
   information from the database. You will be required to re-configure the Ambari server and re-run the cluster wizard.
   Are you SURE you want to perform the reset [yes/no] (no)? yes
   Confirm server reset [yes/no](no)? yes
   

4. Reinstall a supported Hadoop distribution. See Installing a Supported Hadoop Distribution.
5. Restart Trafodion:

   su trafodion
   cd $MY_SQROOT/scripts or cds
   sqstart
   

6. Perform a quick sanity check by creating a table with a couple of records using the Trafodion Conversational Interface (sqlci). For example:

   [trafodion@n001 ~]$ sqlci
   Trafodion Conversational Interface 1.1.0
   (c) Copyright 2015 Hewlett-Packard Development Company, LP.
   >>create table test1 (f1 int, f2 int);
   
   --- SQL operation complete.
   >>insert into test1 values(1,1);
   
   --- 1 row(s) inserted.
   >>insert into test1 values(2,2);
   
   --- 1 row(s) inserted.
   >>select * from test1;
   
   F1            F2
   -----------   -----------
             1             1
             2             2
   
   --- 2 row(s) selected.
   >>get tables;
   
   Tables in Schema TRAFODION.SEABASE
   ==================================
   
   TEST1
   
   --- SQL operation complete.
   >>exit;
   

   Those SQL commands should run successfully.

Uninstalling MapR

NOTE: Trafodion does not currently support MapR. Follow these instructions to remove MapR and install one of the supported Hadoop distributions before installing Trafodion.

The following commands should all be done from the first node of the cluster. Do not run them from a machine that will not be part of the Trafodion cluster. Trafodion must be installed on all nodes that host an HBase RegionServer (that is, where a supported Hadoop distribution is installed). For example, if the Hadoop distribution is installed on nodes n001, n002, and n003, then you must install Trafodion on those same nodes (--nodes "n001 n002 n003").

Follow these steps if you are uninstalling MapR and replacing it with a supported Hadoop distribution before installing Trafodion.

1. If you have not done so already, download the latest tar files from downloads.trafodion.org.
2. Create the trafodion_downloads directory and place the downloaded trafodion-1.1.<n>.tar.gz and installer-1.1.<n>_v<num>.tar.gz tar files into it:

   mkdir $HOME/trafodion_downloads
   mv <your-download-path>/trafodion-1.1.<n>.tar.gz $HOME/trafodion_downloads
   mv <your-download-path>/installer-1.1.<n>_v<num>.tar.gz $HOME/trafodion_downloads
   cd $HOME/trafodion_downloads
   

3. Untar the downloaded files:

   tar -xzf trafodion-1.1.<n>.tar.gz
   tar -xzf installer-1.1.<n>_v<num>.tar.gz
   

4. Run mapr_uninstall:

   NOTE: You must have root access to the machine.

   Required Parameters	Description
   --help	Print this message and exit.
   --nodes "<node_list>"	Space-separated list of nodes (within quotes)

   EXAMPLES:
   cd /root/trafodion_downloads/installer/tools
   ./mapr_uninstall --nodes "redhat-1 redhat-2"
   ./mapr_uninstall
   

5. Install a supported Hadoop distribution. See Installing a Supported Hadoop Distribution.
6. Finish installing Trafodion. See step 5 in Installing Trafodion.

Uninstalling Trafodion and the Hadoop Distribution

These instructions assume that you already have Trafodion and a Hadoop distribution installed on your system. They explain how to use the trafodion_uninstaller to remove Trafodion from your system.

• Uninstalling Trafodion
• Uninstalling Trafodion and Cloudera
• Uninstalling Trafodion and Hortonworks/Ambari

NOTE: If you want to upgrade to a newer version of Trafodion, you do not need to use the trafodion_uninstaller script. Instead, use the trafodion_install script, which automatically upgrades the version of Trafodion. For more information, see Upgrading Trafodion.

Uninstalling Trafodion

The following commands should all be done from the first node of the cluster. Do not run them from a machine that is not part of the Trafodion cluster. Follow these steps only if you are uninstalling Trafodion and not uninstalling the currently running Hadoop distribution.

1. Stop Trafodion:

   su trafodion
   cd $MY_SQROOT/sql/scripts or cds
   sqstop
   

2. Run the trafodion_uninstaller script by providing the following parameters:

   NOTE: To use the --all option, you will need root access to the machine and will need to exit from the trafodion user ID.

   Required Parameters	Description
   --instance <dir-name>	Directory path of the Trafodion instance to remove. Run this option from the trafodion user ID.
   --all	This will remove Trafodion completely, including the trafodion user ID and its /root directory. Run this option from a user ID with sudo access.

   EXAMPLES:
   cd /root/trafodion_downloads/installer
   ./trafodion_uninstaller --instance /home/trafodion/trafodion
   ./trafodion_uninstaller --all
   

Uninstalling Trafodion and Cloudera

The following commands should all be done from the first node of the cluster. Do not run them from a machine that is not part of the Trafodion cluster. Follow these steps if you intend to uninstall both Trafodion and Cloudera.

1. Stop Trafodion:

   su trafodion
   cd $MY_SQROOT/sql/scripts or cds
   sqstop
   

2. Run the trafodion_uninstaller script by providing the following parameters:

   NOTE: To use the --all option, you will need root access to the machine and will need to exit from trafodion user ID.

   Required Parameters	Description
   --instance <dir-name>	Directory path of the Trafodion instance to remove. Run this option from the trafodion user ID.
   --all	This will remove Trafodion completely, including the trafodion user ID and its /root directory. Run this option from a user ID with sudo access.

   EXAMPLES:
   cd /root/trafodion_downloads/installer
   ./trafodion_uninstaller --instance /home/trafodion/trafodion
   ./trafodion_uninstaller --all
   

3. Run cloudera_uninstall:

   NOTE: You must have root access to the machine.

   Required Parameters	Description
   --help	Print this message and exit.
   --nodes "<node_list>"	Space-separated list of nodes (within quotes)

   EXAMPLES:
   cd /root/trafodion_downloads/installer/tools
   ./cloudera_uninstall --nodes "redhat-1 redhat-2"
   

4. To reinstall Trafodion and a different Hadoop distribution, see the Trafodion Installation instructions.

Uninstalling Trafodion and Hortonworks/Ambari

The following commands should all be done from the first node of the cluster. Do not run them from a machine that is not part of the Trafodion cluster. Follow these steps if you intend to uninstall both Trafodion and Hortonworks/Ambari.

1. Stop Trafodion:

   su trafodion
   cd $MY_SQROOT/sql/scripts or cds
   sqstop
   

2. Run the trafodion_uninstaller script by providing the following parameters:

   NOTE: To use the --all option, you will need root access to the machine and will need to exit from trafodion user ID.

   Required Parameters	Description
   --instance <dir-name>	Directory path of the Trafodion instance to remove. Run this option from the trafodion user ID.
   --all	This will remove Trafodion completely, including the trafodion user ID and its /root directory. Run this option from a user ID with sudo access.

   EXAMPLES:
   cd /root/trafodion_downloads/installer
   ./trafodion_uninstaller --instance /home/trafodion/trafodion
   ./trafodion_uninstaller --all
   

3. Run hortonworks_uninstall:

   NOTE: You must have root access to the machine.

   Required Parameters	Description
   --help	Print this message and exit.
   --nodes "<node_list>"	Space-separated list of nodes (within quotes)

   EXAMPLES:
   cd /root/trafodion_downloads/installer/tools
   ./hortonworks_uninstall --nodes "redhat-1 redhat-2"
   

4. During uninstall, the Ambari Server database must be reset. Type 'yes' when prompted.

   EXAMPLE:
   Resetting ambari-server
   **** WARNING **** You are about to reset and clear the Ambari Server database. This will remove all cluster host and configuration 
   information from the database. You will be required to re-configure the Ambari server and re-run the cluster wizard.
   Are you SURE you want to perform the reset [yes/no] (no)? yes
   Confirm server reset [yes/no](no)? yes
   

5. To reinstall Trafodion and a different Hadoop distribution, see the Installing Trafodion on a cluster instructions.