sas visual analytics 7.3 installation and administration guide · sas visual analytics includes the...
TRANSCRIPT
SAS Visual Analytics 7.3 Installation and
Administration Guide
I. Pre-Installation Checklist
II. Password-less SSH for Root
III. HPCMC Installation
IV. Creating Users with Public/Private SSH
keys
V. Hadoop and TKGrid Installation
VI. SAS Visual Analytics Installation
VII. Register LASRadm to Metadata
VIII. Validation
IX. Start-up and Stopping of Services
I. Pre-installation Checklist
Prior to installation, SAS will provide a pre-installation checklist that must be validated before
starting the activity. Installation should not start without completing the checklist. Below is a
sample pre-install checklist for VA 7.3
Operating System Configuration & Patches (Initial)
Requirement: Expected Value:
Red Hat Enterprise Linux 6 (64-Bit version) RHEL 6
Python 2.6
Korn Shell ksh-20100621-6.el6.x86_64
SELINUX Permissive
PERL-NET-SSLEAY PERL-NET-SSLEAY-1.35-9.el6.x86_64 (Blade 1)
X11 Forwarding Enabled
Umask value for sasinst account 22
Ulimit for sasinst user Open Files = (-n) 350000
Max User Processes = (-u) 65536
Installed Libraries & RPMs
numactl-2.0.3-9.el6.x86_64
libXext-1.1-3.el6.x86_64
libXext-1.1-3.el6.i686
libXp-1.0.0-15.1.el6.x86_64
libXp-1.0.0-15.1.el6.i686
libXtst-1.0.99.2-3.el6.x86_64
libXtst-1.0.99.2-3.el6.i686
libgcc 32 & 64 bit versions
libXext-1.1-3.el6.i686
libXp-1.0.0-15.1.el6.i686
libXtst-1.0.99.2-3.el6.i686
Initial Operating System Accounts
root
sasinst
sassrv
sasdemo
Group sasusers
II. Password-less access for root (SSH setup)
The root user account must have password-less secure shell (SSH) access amongst all the
machines in the grid.
0) - Add /etc/hosts
- Turn off firewall
- The same root password
- The same OS version
1) Log in to the SAS node, Nimitz1 and switch to root user.
su -
<password>
2) Generate the SSH keys for root
cd ~
ssh-keygen -t rsa
Enter file in which to save the key (/root/.ssh/id_rsa): /root/.ssh/id_rsa
Enter passphrase (empty for no passphrase):
Enter same passphrase again:
3) Issue the “cat” system command to create the authorized_keys file.
cat ~/.ssh/id_rsa.pub >> ~/.ssh/authorized_keys
4) Verify settings.
ls -al .ssh
a. The ownership of the /root/.ssh dir should be 700 (drwx------)
b. id_rsa: root’s private key. Owned by root with a mode of 600 (-rw-------)
c. id_rsa.pub: root’s public key. Owned by root with a mode of 644 (-rw-r--r--)
You should get a listing similar to the following:
[root@Nimitz1 .ssh]# ls -al .ssh
total 24
drwx------. 2 root root 4096 Dec 30 07:14 .
dr-xr-x---. 26 root root 4096 Dec 30 07:13 ..
-rw-r--r--. 1 root root 398 Dec 30 07:14 authorized_keys
-rw-------. 1 root root 1675 Dec 30 07:14 id_rsa
-rw-r--r--. 1 root root 398 Dec 30 07:14 id_rsa.pub
-rw-r--r--. 1 root root 407 Dec 30 2011 known_hosts
5) Use the scp command to copy the folder /root/.ssh from this server to the other nodes
assigned to the cluster, while maintaining permission and ownership. You are prompted for
the password of the root account each time.
scp -o StrictHostKeyChecking=no -r ~/.ssh/* Stennis2:~/.ssh
<password>
scp -o StrictHostKeyChecking=no -r ~/.ssh/* Vinsone3:~/.ssh
<password>
scp -o StrictHostKeyChecking=no -r ~/.ssh/* Reagan4:~/.ssh
<password>
This step - and the next one - also lay down the server’s public key that is necessary for
connecting without a password.
You should get a series of messages similar to the following:
Warning: Permanently added 'Stennis2,149.173.98.54' (RSA) to the list of known hosts.
root@Stennis2's password:
authorized_keys 100% 398 0.4KB/s 00:00
6) Execute the SSH command to each node on the cluster to test the ssh password-less
connection.
ssh -o StrictHostKeyChecking=no -o PasswordAuthentication=no Nimitz1
exit
ssh -o PasswordAuthentication=no Stennis2
exit
ssh -o PasswordAuthentication=no Vinsone3
exit
ssh -o PasswordAuthentication=no Reagan4
exit
With each of these commands, you are explicitly disabling the use of a password to force
the use of the newly created public keys.
The first line in the sequence opens an SSH connection from Nimitz1 back to itself.
You see the following:
[root@Nimitz1 .ssh]# ssh -o StrictHostKeyChecking=no -o PasswordAuthentication=no
Nimitz1
Warning: Permanently added 'Nimitz1,149.173.96.126' (RSA) to the list of known hosts.
Last login: Mon Dec 12 03:19:28 2012 from Nimitz1
[root@Nimitz1 ~]# exit
logout
Connection to Nimitz1 closed.
III. SAS High-Performance Management Console
SAS Visual Analytics includes the SAS High-Performance Computing Management Console to
handle tasks that are typically performed by a Linux system administrator; customers with Linux
savvy system administrators can choose not to use the SAS HPC MC at all.
This is a completely different application from the SAS Management Console: it does not
connect to or update any metadata in the SAS Metadata Server.
We use it for creating and propagating user accounts and SSH keys across the environment.
The SAS HPC Management Console is installed on the first server and is a web application on
the URL http://Nimitz1:10020
3.1 Installation of HPC Management Console (HPC MC) 1) Log in to the SAS node, Nimitz1, as the sasinst user, and switch to root.
su -
<password>
2) Install the HPC MC with the following commands.
cd /tmp/sasdepot/depot/
cd standalone_installs/SAS_High-Performance_Management_Console/2_1/Linux_for_x64/
rpm -ivh sashpcmc-2.1.x86_64.rpm
You should get the following:
[root@Nimitz1 Linux_for_x64]# rpm -ivh sashpcmc-2.1.x86_64.rpm
Preparing... ##################################### [100%]
1:sashpcmc ##################################### [100%]
3) To configure the HPC MC run the setup script in /opt/webmin/utilbin
cd /opt/webmin/utilbin/
./setup
Provide the following details:
hpcadmin
yes
lnxsas
lnxsas
no
You should get the following:
[root@Nimitz1 utilbin]# ./setup
Enter the username for initial login to SAS HPC MC below. This user will have rights to
everything
in the SAS HPC MC and can either be an OS account or new console user. If an OS
account exists for the user,
then system authentication will be used. If an OS account does not exist, you will be
prompted for a password.
hpcadmin
hpcadmin is not a OS account. Proceed with setting up account as console user only
(yes|no)?
yes
Enter password for user hpcadmin
Confirm password for user hpcadmin
Creating hpcadmin using password
Use SSL\HTTPS (yes|no)?
no
Finished!
4) To test if the setup is successful, start the console with the service1 command:
service sashpcmc status
service sashpcmc start
You should get the following:
[root@Nimitz1 Linux_for_x64]# service sashpcmc status
sashpcmc is stopped
[root@Nimitz1 Linux_for_x64]# service sashpcmc start
sashpcmc started
VERIFICATION STEPS:
1) The HPC MC is installed in /opt/webmin and the whole directory is owned by root. Check it:
1 It’s worth mentioning that the service command is a RHEL feature. If you want to use a command that works on
both SUSE and RHEL, use:
/etc/init.d/sashpcmc status
/etc/init.d/sashpcmc start
ls -l /opt/webmin/
You should get the following:
[root@Nimitz1 utilbin]# ls -l /opt/webmin/
total 20
drwxr-xr-x. 23 root root 4096 Dec 30 07:57 etc
drwxr-xr-x. 2 root root 4096 Dec 30 07:58 logs
drwxr-xr-x. 2 root root 4096 Dec 30 07:57 utilbin
drwxr-xr-x. 3 root root 4096 Dec 30 07:57 var
drwxr-xr-x. 24 root root 4096 Dec 30 07:57 webroot
2) The HPC MC is configured to start automatically when the server starts. Issue the following
command.
chkconfig --list sashpcmc
You should get a listing similar to the following:
[root@Nimitz1 Linux_for_x64]# chkconfig --list sashpcmc
sashpcmc 0:off 1:off 2:on 3:on 4:on 5:on 6:off
3) On the windows client, open a browser and point to the console: http://Nimitz1:10020/
4) Log in to the HPC Management Console with the hpcadmin account, and then log off.
3.2 Creating the gridhosts file with HPCMC 1) On the Windows machine, open Notepad++ to create a .txt file that contains a list of nodes.
You can copy and paste the following list:
Nimitz1
Stennis2
Vinsone3
Reagan4
2) Save and close the file in the /tmp directory (/tmp/gridhosts.txt)
3) Log in to the HPC Management Console with the hpcadmin account, and navigate to the
Generate Gridhosts File utility.
HPC Management Gridhosts File Management Generate Gridhosts File
tab
4) Select “Upload a file” button.
5) On the “Select File to Upload” tab, browse to the local gridhost file that was created in /tmp
6) Select Upload, you get a form with server names listed
7) Press “Select all”, Submit
8) Logoff the SAS HPC MC
VERIFICATION STEP:
1) From the SAS Node, verify that the gridhosts file was created.
cat /etc/gridhosts
You should get the following:
[root@Nimitz1 Linux_for_x64]# cat /etc/gridhosts
Nimitz1
Stennis2
Vinsone3
Reagan4
3.2.1 Simultaneous Utility Commands (simsh and simcp) The simsh and simcp commands are installed with the SAS Visual Analytics software.
The following are some key facts about them:
Typically, executed from the head node.
The /etc/gridhosts file must exist.
The user executing simsh/simcp must have password-less SSH setup properly to every
node in the environment.
Installed with SAS HPC Management Console: /opt/webmin/utilbin
The simsh command uses secure shell to invoke the specified command on every
machine that is listed in the /etc/gridhosts file.
The simsh command can also be used to simplify installation invoking the rpm or
another package installation command on each machine in the cluster.
The simcp command is used to copy a file from one machine to the other machines
listed in /etc/gridhosts file.
1) Log in to the SAS node, Nimitz1, as the sasinst user, and then switch to root.
su -
<password>
2) Test a simultaneous command.
/opt/webmin/utilbin/simsh hostname
You should get the following:
[root@Nimitz1 ~]# /opt/webmin/utilbin/simsh hostname
Vinsone3: Vinsone3
Stennis2: Stennis2
Nimitz1: Nimitz1
Reagan4: Reagan4
Note that the order in which hosts execute the command is not guaranteed.
IV. Creating Users with Public/Private SSH keys
sasinst – SAS installation account;
lasradm – data prep admin to load/lift data
sasdemo - SAS VA end user
sassrv - SAS Spawned Servers account
hadoop - Hadoop user – Starts Hadoop services
4.1 sasinst
This account is used to install;
1) SAS Servers
2) SAS LASR Analytic Server
To install the SAS LASR Analytic Server SSH from any node to any node will be required.
Therefore, the private SSH key and authorized_keys file will be required on all nodes.
In our environment the account already exists on all servers because it is in the pre-installation
checklist; this is to mimic scenarios where consultants do not directly log in to servers with root
account, but rather use the SAS Installation account and then “su” to other ones.
1) From a browser on the windows server, log in to HPCMC as hpcadmin / SASpw123 -
http://Nimitz1:10020/
2) Select HPC Management Users and Groups
3) Select the sasinst user from the table by clicking on the sasinst name within the table
4) Select Generate and Propagate SSH Keys and select Save
4.2 lasradm
This user exists on all nodes of the cluster included the SAS node.
The public and private keys are generated for this user on the SAS node and copied to
all nodes of the environment.
This user must be able to ssh from the SAS node to all nodes on the cluster, as well as
ssh amongst all the nodes in the cluster.
An account is created in the SAS Metadata for this login.
1) From a browser on the windows server, log in to HPCMC as hpcadmin / SASpw123
(http://Nimitz1:10020/)
2) Select HPC Management Users and Groups
3) Select Create a new user and complete with values as per the following screenshot (enter
SASpw123 in the password field)
4) Select Create. You should get a confirmation page for it.
5) Verification Step
Log in to the SAS node, Nimitz1, su to lasradm and submit a remote command to each host:
su - lasradm
<password>
ssh -o StrictHostKeyChecking=no Nimitz1 -C "hostname"
ssh -o StrictHostKeyChecking=no Stennis2 -C "hostname"
ssh -o StrictHostKeyChecking=no Vinsone3 -C "hostname"
ssh -o StrictHostKeyChecking=no Reagan4 -C "hostname"
If correct you do not get a password request:
$ ssh -o StrictHostKeyChecking=no Nimitz1 -C "hostname"
Warning: Permanently added 'Nimitz1,129.8.160.202' (RSA) to the list of known hosts.
Nimitz1
$ ssh -o StrictHostKeyChecking=no Stennis2 -C "hostname"
Warning: Permanently added 'Stennis2,129.8.160.203' (RSA) to the list of known hosts.
Stennis2
$ ssh -o StrictHostKeyChecking=no Vinsone3 -C "hostname"
Warning: Permanently added 'Vinsone3,129.8.160.204' (RSA) to the list of known hosts.
Vinsone3
$ ssh -o StrictHostKeyChecking=no Reagan4 -C "hostname"
Warning: Permanently added 'Reagan4,129.8.160.205' (RSA) to the list of known hosts.
Reagan4.
4.3 sasdemo
Used to log in to the SAS Visual Analytics Web applications.
Only needed on the SAS node, is not needed on the cluster nodes.
An account is created in the SAS Metadata for this login.
No ssh keys are created on any of the nodes.
In our environment the account already exists on all servers because it is in the pre-installation
checklist;
1) Log in to the SAS node, Nimitz1, and su to root user
su –
<password>
2) Use the simsh command to simultaneously check the uid:
/opt/webmin/utilbin/simsh cat /etc/passwd | grep sasdemo
You should get a listing similar to the following :
[root@Nimitz1 ~]# /opt/webmin/utilbin/simsh cat /etc/passwd | grep sasdemo
Nimitz1: sasdemo:x:503:504::/home/sasdemo:/bin/bash
Vinsone3: sasdemo:x:503:504::/home/sasdemo:/bin/bash
Stennis2: sasdemo:x:503:504::/home/sasdemo:/bin/bash
Reagan4: sasdemo:x:503:504::/home/sasdemo:/bin/bash
3) Use the simsh command to simultaneously verify that sasdemo has no ssh keys set up.
/opt/webmin/utilbin/simsh ls -la /home/sasdemo/.ssh
You should get a listing similar to the following:
[root@Nimitz1 ~]# /opt/webmin/utilbin/simsh ls -la /home/sasdemo/.ssh
ls: cannot access /home/sasdemo/.ssh: No such file or directory
ls: cannot access /home/sasdemo/.ssh: No such file or directory
ls: cannot access /home/sasdemo/.ssh: No such file or directory
ls: cannot access /home/sasdemo/.ssh: No such file or directory
4.4 sassrv
Shared login used by standard BI servers.
Only needed on the SAS node, is not needed on the cluster nodes.
An account is created in the SAS Metadata for this login.
No ssh keys are created on any of the nodes.
In our environment the account already exists on all servers because it is in the pre-installation
checklist; this is to mimic scenarios where consultants do not directly log in to servers with root
account, but rather use the SAS Installation account and.
1) Log in to the SAS node, Nimitz1, and su to root user
su –
<password>
2) Use the simsh command to simultaneously check the uid:
/opt/webmin/utilbin/simsh cat /etc/passwd | grep sassrv
You should get a listing similar to the following:
[root@Nimitz1 ~]# /opt/webmin/utilbin/simsh cat /etc/passwd | grep sassrv
Nimitz1: sassrv:x:505:504::/home/sassrv:/bin/bash
Stennis2: sassrv:x:505:504::/home/sassrv:/bin/bash
Reagan4: sassrv:x:505:504::/home/sassrv:/bin/bash
Vinsone3: sassrv:x:505:504::/home/sassrv:/bin/bash
3) Use the simsh command to simultaneously verify that sassrv has no ssh keys set up.
/opt/webmin/utilbin/simsh ls -la /home/sassrv/.ssh
You should get a listing similar to the following:
[root@Nimitz1 ~]# /opt/webmin/utilbin/simsh ls -la /home/sassrv/.ssh
ls: cannot access /home/sassrv/.ssh: No such file or directory
ls: cannot access /home/sassrv/.ssh: No such file or directory
ls: cannot access /home/sassrv/.ssh: No such file or directory
ls: cannot access /home/sassrv/.ssh: No such file or directory
4.5 hadoop
The account used to run the Hadoop Distributed File System (HDFS) will be hadoop. The
software will be both installed and run using this user account. The Hadoop user requires SSH
from any node to any node to complete the installation. The installer does not create this
account automatically and it must be created manually using the HPCMC.
1) From a browser on the windows server, log in to HPCMC as hpcadmin / SASpw123 -
http://Nimitz1:10020/
2) Select HPC Management Users and Groups
3) Select Create a new user and complete with values as per the following screenshot (enter
SASpw123 in the password field)
4) Select Create. You should get a confirmation page.
VERIFICATION STEPS
[root@Nimitz1 ~]# su – Hadoop
ssh -o StrictHostKeyChecking=no -o PasswordAuthentication=no Nimitz1
exit
ssh -o PasswordAuthentication=no Stennis2
exit
ssh -o PasswordAuthentication=no Reagan4
exit
ssh -o PasswordAuthentication=no Vinsone3
exit
V. Hadoop and TKGrid Installation
5.1 Create the Installation Directory
Login to the Hadoop Name node, Nimitz1, as the sasinst user and change to root
su -
<password>
Use the simsh script to create the hadoop installation directory on all nodes
Note : /Hadoop should be in the large drive
Example : /data/Hadoop
Default Hadoop data is : /Hadoop/Hadoop-data
/opt/webmin/utilbin/simsh mkdir /hadoop
/opt/webmin/utilbin/simsh chown hadoop:hadoop /hadoop
5.2 Install the SAS Visual Analytics Hadoop Software
Login to the Hadoop Name node, Nimitz1, as the root user, then change to the hadoop
user, use the following command as root
su - hadoop
cd ~
Copy the SAS High-Performance Deployment for Hadoop installer to the hadoop user’s
home directory
cp /tmp/sasdepot/depot/standalone_installs/SAS_High-
Performance_Deployment_for_Hadoop/2_1/Linux_for_x64/sashadoop.tar.gz .
Extract the installer
tar -xvf sashadoop.tar.gz
Run the installation script
cd sashadoop
./hadoopInstall
Use the values in the following table to install SAS High-Performance Deployment for
Hadoop
Question Response
Do you wish to use an existing Hadoop installation (y/N) N
Enter path to install Hadooop. The directory ‘hadoop-0.23.0’ will be created in
the path specified
/hadoop
Enter replication factor. Default 2 2
Enter port number for fs.defaultFS. Default 54310 <press enter>
Enter port number for mapred.job.tracker. Default 54311 <press enter>
Enter port number for dfs.namenode.https-address. Default 50470 <press enter>
Enter port number for dfs.datanode.https.address. Default 50475 <press enter>
Enter port number for dfs.datanode.address. Default 50010 <press enter>
Enter port number for dfs.datanode.ipc.address. Default 50020 <press enter>
Enter port number for dfs.namenode.http-address. Default 50070 <press enter>
Enter port number for dfs.datanode.http.address. Default 50075 <press enter>
Enter port number for dfs.secondary.http.address. Default 50090 <press enter>
Enter port number for dfs.namenode.backup.address. Default 50100 <press enter>
Enter port number for dfs.namenode.backup.http-address. Default 50105 <press enter>
Enter port number for mapred.job.tracker.http.address. Default 50030 <press enter>
Enter port number for mapred.task.tracker.http.address. Default 50060 <press enter>
Enter port number for com.sas.lasr.hadoop.service.namenode.port. Default
15452
<press enter>
Enter port number for com.sas.lasr.hadoop.service.datanode.port. Default
15453
<press enter>
Enter user that will be running the HDFS server process. hadoop
Enter path for JAVA_HOME directory. (Default: /usr/lib/jvm/jre) <press enter>
Enter path for Hadoop data directory. This should be on a large drive.
Default is '/hadoop/hadoop-data'.
<press enter>
Enter path for Hadoop system directory. Default is '/hadoop/hadoop-system'. <press enter>
Enter path for Hadoop local directory. Default is '/hadoop/hadoop-local'. <press enter>
Enter path for Hadoop name directory. Default is '/hadoop/hadoop-name'. <press enter>
Enter full path to machine list. The NameNode 'Nimitz1' should be listed first. /etc/gridhosts
You should see the following:
Creating 'hadoop-0.23.1' in '/hadoop'.
When the TKGrid installer prompts for 'HADOOP_HOME', enter '/hadoop/hadoop-0.23.1'.
Copying SAS jars to '/hadoop/hadoop-0.23.1/share/hadoop/hdfs/lib'.
Creating Hadoop tmp dir in '/hadoop/hadoop-0.23.1/tmp'.
Creating '/hadoop/hadoop-0.23.1/etc/hadoop/masters' file.
Creating '/hadoop/hadoop-0.23.1/etc/hadoop/slaves' file.
Adding SAS settings to '/hadoop/hadoop-0.23.1/etc/hadoop/hdfs-site.xml'.
mkdir -m 755 /hadoop/hadoop-local /hadoop/hadoop-data /hadoop/hadoop-name
/hadoop/hadoop-system
chown hadoop /hadoop/hadoop-local /hadoop/hadoop-data /hadoop/hadoop-name
/hadoop/hadoop-system
mkdir -m 777 /hadoop/hadoop-0.23.1/logs
touch /hadoop/hadoop-0.23.1/logs/SecurityAuth.audit
chmod 777 /hadoop/hadoop-0.23.1/logs/SecurityAuth.audit
The installer can now copy '/hadoop/hadoop-0.23.1' to all the slave machines
using scp, skipping the first entry. Perform copy? (YES/no)
Answer “YES” and you should see the following as the files are copied to all nodes:
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Stennis2 mkdir -p -m
755 /hadoop/hadoop-0.23.1
scp -o StrictHostKeyChecking=no -o PasswordAuthentication=no -q -r /hadoop/hadoop-
0.23.1/* Stennis2:/hadoop/hadoop-0.23.1/
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Stennis2 chmod -R 755
/hadoop/hadoop-0.23.1
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Stennis2 chmod -R 777
/hadoop/hadoop-0.23.1/logs /hadoop/hadoop-0.23.1/tmp
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Stennis2 mkdir -m 755
/hadoop/hadoop-local /hadoop/hadoop-data /hadoop/hadoop-name /hadoop/hadoop-system
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Stennis2 chown hadoop
/hadoop/hadoop-local /hadoop/hadoop-data /hadoop/hadoop-name /hadoop/hadoop-system
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Reagan4 mkdir -p -m
755 /hadoop/hadoop-0.23.1
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Vinsone3 mkdir -p -m
755 /hadoop/hadoop-0.23.1
scp -o StrictHostKeyChecking=no -o PasswordAuthentication=no -q -r /hadoop/hadoop-
0.23.1/* Reagan4:/hadoop/hadoop-0.23.1/
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Stennis2 scp -o
StrictHostKeyChecking=no -o PasswordAuthentication=no -q -r /hadoop/hadoop-0.23.1/*
Vinsone3:/hadoop/hadoop-0.23.1/
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Vinsone3 chmod -R
755 /hadoop/hadoop-0.23.1
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Reagan4 chmod -R 755
/hadoop/hadoop-0.23.1
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Vinsone3 chmod -R
777 /hadoop/hadoop-0.23.1/logs /hadoop/hadoop-0.23.1/tmp
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Reagan4 chmod -R 777
/hadoop/hadoop-0.23.1/logs /hadoop/hadoop-0.23.1/tmp
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Vinsone3 mkdir -m 755
/hadoop/hadoop-local /hadoop/hadoop-data /hadoop/hadoop-name /hadoop/hadoop-system
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Reagan4 mkdir -m 755
/hadoop/hadoop-local /hadoop/hadoop-data /hadoop/hadoop-name /hadoop/hadoop-system
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Vinsone3 chown
hadoop /hadoop/hadoop-local /hadoop/hadoop-data /hadoop/hadoop-name
/hadoop/hadoop-system
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Reagan4 chown
hadoop /hadoop/hadoop-local /hadoop/hadoop-data /hadoop/hadoop-name
/hadoop/hadoop-system
Installation complete.
Please add the following line to /etc/sudoers on all nodes:
hadoop ALL=NOPASSWD:/bin/mkdir,/bin/chmod,/bin/chown,/bin/ln,/bin/rm
Start your HDFS as userid 'hadoop' with the following commands:
export HADOOP_HOME=/hadoop/hadoop-0.23.1
Run this command only if you want to format a new hadoop directory. It will erase existing
data.
$HADOOP_HOME/bin/hadoop namenode -format
$HADOOP_HOME/sbin/start-dfs.sh
VERIFICATION STEP:
Verify hadoop directory creation on all nodes:
/opt/webmin/utilbin/simsh ls -la /hadoop
You should get a listing similar to the following:
[hadoop@Nimitz1 sashadoop]$ /opt/webmin/utilbin/simsh ls -la /hadoop
Nimitz1: drwxr-xr-x. 2 hadoop hadoop 4096 Oct 30 11:24 hadoop-system
Nimitz1: drwxr-xr-x. 2 hadoop hadoop 4096 Oct 30 11:24 hadoop-name
Nimitz1: drwxr-xr-x. 2 hadoop hadoop 4096 Oct 30 11:24 hadoop-local
Nimitz1: drwxr-xr-x. 2 hadoop hadoop 4096 Oct 30 11:24 hadoop-data
Nimitz1: drwxr-xr-x. 11 hadoop hadoop 4096 Oct 30 11:24 hadoop-0.23.1
Nimitz1: dr-xr-xr-x. 29 root root 4096 Oct 30 11:07 ..
Nimitz1: drwxr-xr-x. 7 hadoop hadoop 4096 Oct 30 11:24 .
Nimitz1: total 28
Nimitz1: Warning: Permanently added 'Nimitz1,10.96.3.64' (RSA) to the list of known hosts.
Stennis2: drwxr-xr-x. 2 hadoop hadoop 4096 Oct 30 11:25 hadoop-system
Stennis2: drwxr-xr-x. 2 hadoop hadoop 4096 Oct 30 11:25 hadoop-name
Stennis2: drwxr-xr-x. 2 hadoop hadoop 4096 Oct 30 11:25 hadoop-local
Stennis2: drwxr-xr-x. 2 hadoop hadoop 4096 Oct 30 11:25 hadoop-data
Stennis2: drwxr-xr-x. 11 hadoop hadoop 4096 Oct 30 11:25 hadoop-0.23.1
Stennis2: dr-xr-xr-x. 29 root root 4096 Oct 30 11:07 ..
Stennis2: drwxr-xr-x. 7 hadoop hadoop 4096 Oct 30 11:25 .
Stennis2: total 28
Vinsone3: drwxr-xr-x. 2 hadoop hadoop 4096 Oct 30 11:26 hadoop-system
Vinsone3: drwxr-xr-x. 2 hadoop hadoop 4096 Oct 30 11:26 hadoop-name
Vinsone3: drwxr-xr-x. 2 hadoop hadoop 4096 Oct 30 11:26 hadoop-local
Vinsone3: drwxr-xr-x. 2 hadoop hadoop 4096 Oct 30 11:26 hadoop-data
Vinsone3: drwxr-xr-x. 11 hadoop hadoop 4096 Oct 30 11:26 hadoop-0.23.1
Vinsone3: dr-xr-xr-x. 29 root root 4096 Oct 30 11:07 ..
Vinsone3: drwxr-xr-x. 7 hadoop hadoop 4096 Oct 30 11:26 .
Vinsone3: total 28
Reagan4: drwxr-xr-x. 2 hadoop hadoop 4096 Oct 30 11:26 hadoop-system
Reagan4: drwxr-xr-x. 2 hadoop hadoop 4096 Oct 30 11:26 hadoop-name
Reagan4: drwxr-xr-x. 2 hadoop hadoop 4096 Oct 30 11:26 hadoop-local
Reagan4: drwxr-xr-x. 2 hadoop hadoop 4096 Oct 30 11:26 hadoop-data
Reagan4: drwxr-xr-x. 11 hadoop hadoop 4096 Oct 30 11:26 hadoop-0.23.1
Reagan4: dr-xr-xr-x. 29 root root 4096 Oct 30 11:07 ..
Reagan4: drwxr-xr-x. 7 hadoop hadoop 4096 Oct 30 11:26 .
Reagan4: total 28
Change to the root user
su -
<password>
Update the /etc/sudoers file on each node, edit the file using the following command:
visudo
Add the following two lines to the bottom of the file:
Defaults:hadoop !requiretty
hadoop ALL=NOPASSWD:/bin/mkdir,/bin/chmod,/bin/chown,/bin/ln,/bin/rm
Save and exit the file. Use the following command to copy the file to all nodes:
/opt/webmin/utilbin/simcp /etc/sudoers /etc/sudoers
Change to the hadoop user
su - hadoop
Update the .bash_profile for the hadoop user to set the environment variable
$HADOOP_HOME
echo "export HADOOP_HOME=/hadoop/hadoop-2.4.0" >> ~/.bash_profile
Process the bash_profile to make the environment variable available
. ~/.bash_profile
Test :
echo $HADOOP_HOME
/hadoop/hadoop-2.4.0
Format the Hadoop file system using the following command:
$HADOOP_HOME/bin/hdfs namenode -format
A large amount of output will be sent to the screen following the format command.
Make sure you see the message: " INFO namenode.NNStorage: Storage directory
/hadoop/hadoop-name has been successfully formatted", a few lines up from the
bottom.
Start the Hadoop file system
$HADOOP_HOME/sbin/start-dfs.sh
You should get some “starting” messages that report the startup of daemons and the
creation of log files. The actual number depends on the number of nodes in the cluster, so at
customer sites it may vary.
Also you will be prompted to continue connecting (yes/no) during the startup of the
processes. Always answer “yes”.
In this workshop environment there should be four “starting” lines. You can also verify that
the correct number of processes (nine) has started on all hosts with the following command:
/opt/webmin/utilbin/simsh ps -ef | egrep "(Node|Tracker)" | egrep -v grep | wc -l
You should get a listing similar to the following:
[hadoop@Nimitz1 ~]$ simsh ps -ef | egrep "(Node|Tracker)" | egrep -v grep | wc -l
5
Still using the hadoop user account, create a directory in HDFS that permits Read and
Write access for all users:
$HADOOP_HOME/bin/hadoop fs -mkdir /hps
$HADOOP_HOME/bin/hadoop fs -chmod 777 /hps
VERIFICATION STEPS:
Validate the creation of your HDFS directory using:
$HADOOP_HOME/bin/hadoop fs -ls -R /
You should get a listing similar to the following:
drwxrwxrwx - hadoop supergroup 0 2012-05-24 20:04 /hps
You can confirm that Hadoop is running successfully by opening a browser to
http:/Nimitz1:50070/dfshealth.jsp. Review the information in the cluster summary section of the
page. Confirm that the number of live nodes equals the number of DataNodes and that the
number of dead nodes is zero.
*In some cases, you might see 0 live and dead nodes. This is due to the firewall settings.
Turn off the firewall/iptables and refresh the page to confirm the live nodes are seen.
Close the browser and exit from the hadoop user session.
exit
5.3 Install the SAS High-Performance Analytics components
Login to the LASR Head node, Nimitz1, as the sasinst user, then switch to root
su -
<password>
Create the installation directory /opt/TKGrid_2.1, on all nodes
/opt/webmin/utilbin/simsh mkdir /opt/TKGrid_2.1
/opt/webmin/utilbin/simsh chown sasinst:sas /opt/TKGrid_2.1
Change to the sasinst user
su - sasinst
Copy the installation file from the depot to the installation directory
cd /opt/TKGrid_2.1/
cp /tmp/sasdepot/depot/standalone_installs/SAS_High-
Performance_Node_Installation/2_1/Linux_for_x64/TKGrid_Linux_x86_64.sh .
Install SAS High-Performance Analytics components, by running the command:
./TKGrid_Linux_x86_64.sh
You will see the following output
Verifying archive integrity... All good.
Uncompressing
TKGrid.......................................................................................................................................
...................................................................................................................................................
...................................................................................................................................................
...................................................................................................................................................
...................................................................................................................................................
...................................................................................................................................................
...................................................................................................................................................
...................................................................................................................................................
...................................................................................................................................................
...................................................................................................................................................
...................................................................................................................................................
...................................................................................................................................................
...................................................................................................................................................
.............................................................
running TKGrid/bin/tkgridconfig TKGrid /opt/TKGrid_2.1
TKGrid Configuration Utility.
Running on 'Nimitz1'
Using stdin for options.
Shared install or replicate to each node? (Y=SHARED/n=replicated
Respond to the prompts from the “TKGrid Configuration Utility”:
Parameter Description
Shared install or replicate to
each node?
(Y=SHARED/n=replicated)
n
Enter additional paths to
include in
LD_LIBRARY_PATH,
separated by colons (:)
Press ENTER. Do not add additional options.
Enter additional options to
mpirun.
Press ENTER. Do not add additional options.
Enter path to use for Utility
files. (default is /tmp).
SAS High-Performance Analytics applications might write scratch
files. By default, these files are created in the /tmp directory.
Press ENTER to accept the default.
Enter path to Hadoop. (default
is Hadoop not installed).
Enter /hadoop/hadoop-0.23.1 unless you installed Hadoop in a
different location.
Force Root Rank to run on
headnode? (y/N)
N
Enter full path to machine list... Enter the name of the file that you created earlier, /etc/gridhosts .
Enter maximum run time for
grid jobs (in seconds). Default
7200 (2 hours).
If a SAS High-Performance Analytics application executes for more
than the maximum allowable run time, it is automatically terminated.
31560000 means 1 year, or 604800 is 1 week.
Keep the default
Enter value for UMASK.
(default is unset.)
Type 022 for the umask value and press ENTER.
You will see the following prompt after answering the prompts above:
The install can now copy this directory to all the machines
listed in '/opt/TKGrid_2.1/TKGrid/grid.hosts' using scp, skipping the first entry. Perform
copy? (YES/no)
Answer YES
The following will be displayed as the installation is copied to each node:
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Stennis2 mkdir -p -m
755 /opt/TKGrid_2.1
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Stennis2 mkdir -m 755
/opt/TKGrid_2.1/TKGrid
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Stennis2 ln -s
bin/tkmpinodelib.sh /opt/TKGrid_2.1/TKGrid/tkmpinodelib.sh
scp -o StrictHostKeyChecking=no -o PasswordAuthentication=no -q -r
/opt/TKGrid_2.1/TKGrid/* Stennis2:/opt/TKGrid_2.1/TKGrid/
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Stennis2 chmod -R 755
/opt/TKGrid_2.1/TKGrid
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Reagan4 mkdir -p -m
755 /opt/TKGrid_2.1
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Vinsone3 mkdir -p -m
755 /opt/TKGrid_2.1
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Reagan4 mkdir -m 755
/opt/TKGrid_2.1/TKGrid
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Vinsone3 mkdir -m 755
/opt/TKGrid_2.1/TKGrid
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Reagan4 ln -s
bin/tkmpinodelib.sh /opt/TKGrid_2.1/TKGrid/tkmpinodelib.sh
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Vinsone3 ln -s
bin/tkmpinodelib.sh /opt/TKGrid_2.1/TKGrid/tkmpinodelib.sh
scp -o StrictHostKeyChecking=no -o PasswordAuthentication=no -q -r
/opt/TKGrid_2.1/TKGrid/* Reagan4:/opt/TKGrid_2.1/TKGrid/
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Stennis2 scp -o
StrictHostKeyChecking=no -o PasswordAuthentication=no -q -r /opt/TKGrid_2.1/TKGrid/*
Vinsone3:/opt/TKGrid_2.1/TKGrid/
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Reagan4 chmod -R 755
/opt/TKGrid_2.1/TKGrid
ssh -q -o StrictHostKeyChecking=no -o PasswordAuthentication=no Vinsone3 chmod -R
755 /opt/TKGrid_2.1/TKGrid
Create a symbolic link /opt/TKGrid pointing to the new installation, first change to root
and then run the command:
su -
<password>
/opt/webmin/utilbin/simsh ln -s /opt/TKGrid_2.1/TKGrid /opt/TKGrid
5.4 Test the HPA Server Cluster Installation
Login to the LASR Head node, Nimitz1, as the sasinst user
Execute the Grid Monitor shell script
/opt/TKGrid/bin/gridmon.sh
As a result, you should have the following window showing up on the Windows machine
VI. SAS Visual Analytics Installation
The following steps show the installation steps required to install SAS Visual Analytics. The plan
file will be provided by SAS. All of the steps above must be completed prior to this installation
step. Most of the values required are default, so there will be minimal inputs required. Please be
guided by the screenshots below.
VII.
VII. Register LASRadm to Metadata
Execute management console and register LASRadm by following the steps below:
Create a New User by right-clicking User Manager and select create a new user
Type lasradm as the account name
From the Groups and Roles tab, select Visual Analytics Data and Visual Data Builder Administrators. Assign lasradm to that group.
Assign lasradm account as the DefaultAuth login. Click ok to finish.
VIII. Validation
You can perform validation by following the steps noted in the Intructions.html file, which is the
result of the SAS Visual Analytics Installation. Below are screenshots of some the steps:
Validation of the SAS Workspace, Pooled Workspace, and Stored Process Servers;
Acessing the VA Hub:
Loading to Hadoop:
Starting the SAS LASR Analytic Server:
Loading a Table to LASR:
IX. Start-up and Stopping SAS Services
To start SAS services on Nimitz1, you must follow the correct sequence as noted below:
1) Start Hadoop (Must be logged in as hadoop user)
$HADOOP_HOME/sbin/start-dfs.sh
2) Start the SAS services (all commands below must be executed as sasinst user)
/sas93/config/Lev1/sas.servers start
3) Start the Information Retrieval Studio
/sas93/config/Lev1/Applications/SASInformationRetrievalStudio*/IRStudio.sh start
4) Start the LASR Monitor Server
/sas93/config/Lev1/Applications/SASVisual*/H*/LASRMonitor.sh start
To stop SAS services on Nimitz1, you must follow the correct sequence as noted below:
1) Stop the LASR Monitor Server (You must use sasinst account for steps 1 – 3)
/sas93/config/Lev1/Applications/SASVisual*/H*/LASRMonitor.sh stop
2) Stop the Information Retrieval Studio
/sas93/config/Lev1/Applications/SASInformationRetrievalStudio*/IRStudio.sh stop
3) Stop the SAS services
/sas93/config/Lev1/sas.servers stop
4) Stop Hadoop Nodes(Must be logged in as hadoop user)
$HADOOP_HOME/sbin/stop-dfs.sh